Applied Energy 331 (2023) 120366 Contents lists available at ScienceDirect Applied Energy journal homepage: www.elsevier.com/locate/apenergy An energy future beyond climate neutrality: Comprehensive evaluations of transition pathways Nick Martin a, Laura Talens-Peiró a, Gara Villalba-Méndez a, b, Rafael Nebot-Medina c, Cristina Madrid-López a, * a Sostenipra Research Group, Institute of Environmental Science and Technology (ICTA-UAB), Autonomous University of Barcelona, Bellaterra (Cerdanyola del Vallès), Catalunya 08193, Spain b Department of Chemical, Biological and Environmental Engineering, Autonomous University of Barcelona, Bellaterra (Cerdanyola del Vallès), Catalunya 08193, Spain c Scientific and Technological Computing Department, Instituto Tecnológico de Canarias, Las Palmas de Gran Canaria, Canarias 35003, Spain H I G H L I G H T S G R A P H I C A L A B S T R A C T • Energy models generally fail to consider impacts beyond direct GHG emissions. • ENBIOS joins a push to integrate life cycle thinking into energy modelling frameworks. • Extra dimensions are added via social metabolism and hierarchical analysis. • Analysis of EU scenarios reveals poten­ tial land use, labour and material issues. A R T I C L E I N F O A B S T R A C T Keywords: Many of the long-term policy decisions surrounding the sustainable energy transition rely on models that fail to Sustainable energy transition consider environmental impacts and constraints beyond direct greenhouse gas emissions and land occupation. Renewable energy Such assessments offer incomplete and potentially misleading information about the true sustainability issues of Energy modelling transition pathways. Meanwhile, although decision-makers desire greater access to a broader range of envi­ Integrated assessment Life cycle assessment ronmental, material and socio-economic indicators, few tools currently address this gap. Here, we introduce Critical raw materials ENBIOS, a framework that integrates a broader range of such indicators into energy modelling and policymaking practices. By calculating sustainability-related indicators across hierarchical levels, we reach deeper un­ derstandings of the potential energy systems to be derived. With ENBIOS, we analyse a series of energy pathways designed by the Calliope energy system optimization model for the European energy system in 2030 and 2050. Although overall emissions will drop significantly, considerable rises in land, labour and critical raw material requirements are likely. These outcomes are further reflected in unfavourable shifts in key metabolic indicators during this period; energy metabolic rate of the system will drop by 25.6%, land requirement-to-energy will quadruple, while the critical raw material supply risk-to-energy ratio will rise by 74.2%. Heat from biomass and electricity from wind and solar are shown to be the dominant future processes across most indicator categories. * Corresponding author. E-mail address: cristina.madrid@uab.cat (C. Madrid-López). https://doi.org/10.1016/j.apenergy.2022.120366 Received 16 August 2022; Received in revised form 26 October 2022; Accepted 14 November 2022 Available online 28 November 2022 0306-2619/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). N. Martin et al. Applied Energy 331 (2023) 120366 Nomenclature ISIC International Standard Industrial Classification kg kilograms Glossary kgCO2-eq kilograms of carbon dioxide equivalent ALOP agricultural land occupation LCIA method km2 square kilometres CRM critical raw material LCA life cycle assessment EC European Commission LCI life cycle inventory EJ exajoules (x1018 joules) LCIA life cycle impact assessment EMR energy metabolic rate (energy per hour of human activity) m2 square metres [MJ/h] MJ megajoules (x106 joules) ENBIOS ENvironmental and BIOeconomic System Assessment MuSIASEM Multi-Scale Integrated Analysis of Socio-Ecosystem EoLRIR end-of-life recycling input rate [%] Metabolism ESM energy system model MW megawatts (x106 watts) EU European Union PgCO2-eq x1012 kilograms of carbon dioxide equivalent Gh gigahours (x109 hours) pLCA prospective life cycle assessment GHG greenhouse gas RoW rest of the world GHGMR GHG metabolic rate (GHG per hour of human activity) SENTINEL Sustainable Energy Transitions Laboratory [kgCO2-eq/h] solar PV solar photovoltaic GWP global warming potential SR supply risk [yr] GWP100 global warming potential (100-year time horizon) LCIA TL teralitres (x1012 litres or 109 cubic metres) method TWh terawatt-hours (x1012 watt-hours) h hours ULOP urban land occupation LCIA method HA human activity [h] WDP water depletion LCIA method IAM integrated assessment model WMR water metabolic rate (water use per hour of human ILO International Labour Organization activity) [m3/h] IPCC Intergovernmental Panel on Climate Change yr years 1. Introduction this life cycle inventory (LCI) data can be converted into more tangible indicator outputs using predefined life cycle impact assessment (LCIA) Global and local responses to the threat of climate change call for methods or other calculations. large reductions in the production of greenhouse gases (GHGs) via a LCA approaches have been widely adopted in recent years for rapid transition towards more sustainable energy system configurations. comparing different existing and emerging technologies, including those In the European Union (EU), the decarbonization of the economy–where within the energy sector [9,10]. Furthermore, they are beginning to be renewable energy technologies replace older fossil fuel-based tech­ used to quantify impacts within complete energy systems by aggregating nologies–acts as one of the five “dimensions” of the energy union the impacts of the various technological processes that occur within strategy [1]. However, while “decarbonised” energy systems are them [11]. Attempts have also been made to integrate LCA data sources generally seen to represent cleaner and more sustainable alternatives, directly into energy models [8,12,13], and by creating simplified in­ wider understandings of the impacts and constraints that relate to ventories on-the-fly within IAM simulations [14]. Conversely, within the different energy technologies are often overlooked in decision-making. burgeoning field of prospective life cycle assessment (pLCA) [15–20] Indeed, the integrated assessment models (IAMs) and other energy information from IAMs, input–output models and other sources system models (ESMs) used to inform energy-related decisions in the EU regarding future variations in energy system configurations–e.g., energy and elsewhere tend to only include simplified estimates of direct GHG mixes or efficiencies–are used to incorporate provisions for future emissions and obviate other environmental factors–such as raw material changes within LCA calculations. or water use– that may limit calculations [2,3]. In any event, no existing approaches allow for the detailed inclusion Incorporating detailed sets of environmental calculations into such and analysis of LCA-related inputs or outputs across hierarchical levels models is cumbersome not only from a computational point of view; it is within energy systems. In this regard, a number of studies have applied also complicated by the vastly different semantics employed to define relational analysis principles to assessing the social metabolism aspects energy and environmental dynamics [4]. Nevertheless, continuing to of energy systems [21,22]. However, such analyses have not attempted overlook or simplify certain environmental aspects and constraints to assess environmental impacts or constraints in any detail, nor have could result in suboptimal outcomes in proposed transition pathways they included other socio-metabolic indicators or been able to integrate [5]. What’s more, research suggests that energy decision makers are the resource needs of those systems. eager to access more comprehensive information about the range of To bridge this gap, we introduce the ENvironmental and BIO­ possible environmental impacts and other limitations that could affect economic System Assessment (ENBIOS) framework. ENBIOS has been future energy scenarios [6]. developed to perform sustainability assessments within the energy The use of life cycle assessment (LCA) [7] approaches is one way to modelling platform developed as part of the Sustainable Energy Tran­ address these shortcomings. Within an LCA analysis, the full life cycle of sitions Laboratory (SENTINEL) project [23]. It connects LCA function­ an energy–or any other–process is considered [8]. So, rather than basing ality with the relational capabilities of the Multi-Scale Integrated quantifications solely on the most obvious or direct aspects of a process, Analysis of Societal and Ecosystem Metabolism (MuSIASEM) approach LCA-based approaches require the collection of input–output in­ [24–27]. ENBIOS takes system definition information from ESMs –or ventories for all contributing operations, from material extraction any other real or theoretical systems– and combines this with built-in through to end-of-life disposal or recycling stages. Input flows include datasets to generate a range of environmental and other indicators at aspects like raw material, land, water and energy use, while output flows each element in that system. This fundamentally includes the use of include emissions to the atmosphere, water bodies and other ecosystems LCA-based datasets and methods. However, any number of other user- as well as the products and co-products of a technology. Collectively, defined methodologies or datasets can be included in these calculations. 2 N. Martin et al. Applied Energy 331 (2023) 120366 Two broad types of indicators are produced within an ENBIOS thinking with energy system configurations and to consider the socio- simulation. The first of these, extensive indicators, are derived using life metabolic dynamics of energy systems, to the best of our knowledge cycle inventory data and other inputs, such as labor requirements, and ENBIOS represents the first attempt to consolidate these two perspec­ provide information about the total size and impacts of the system’s tives into a single framework. To do this, ENBIOS integrates the high- components. For example, total energy production, GHG emissions, land resolution impact assessment capabilities of LCA with the systemic requirements or human labour could all be calculated for a given tech­ upscaling capabilities of MuSIASEM. A summary of the ENBIOS work­ nological process in a system. Combining extensive indicator values for flow is represented in Fig. 1 and detailed in the following sections. each of these elements then allows a second set of intensive indicators to be calculated within the system. Using the previous examples, indicators could be derived to represent the energy produced per hour of labour or 2.1. Preparation GHG emissions per unit of energy, providing useful information about system performance and relations. The first step in the development of an ENBIOS simulation is typi­ The set of extensive and intensive indicators can then be examined cally to define the system framework within a MuSIASEM environment. within and across hierarchical levels using multi-scale system analysis. To do this one must first define a “dendrogram”, a multi-level structure This can provide valuable information about the nature of systems at that arranges the system hierarchically into “processor” nodes where the different levels, from individual processes or grouped categories to relationships between input and output flows are calculated. Processors entire energy systems. For example, visualisations of extensive data can operate in one of two capacities within the “tree” of the dendrogram. could help to identify if higher land requirements are caused by heat or “Structural processors” represent the most specific and tangible activ­ electricity sources in a given scenario, or the relative contributions to ities that can easily be located within a spatial–temporal context (e.g., labour requirements from different technologies across different sce­ specific technologies like wind turbines or nuclear stations). Meanwhile, narios. Likewise, changes in intensive indicators–e.g., a drop in water “functional processors” represent a less tangible social function (e.g., use per unit of energy–can be compared and traced according to the wind turbines or solar PV panels could exist as “structures” related to the changes that occur at different hierarchical levels. “function” of renewable electricity supply, which itself related to elec­ Ultimately, the innovation of the framework is rooted in its ability to tricity supply and, ultimately, energy supply). In other words, the integrate LCA and MuSIASEM methods for deriving extensive and MuSIASEM usage within ENBIOS, the lower levels of the dendrogram intensive indicators to evaluate systems from a hierarchical analysis are typically represented by structures that are later related to the func­ perspective. The flexibility of the framework to different systems and tions they can provide further “up” the hierarchical structure, respecting applications allows simple but powerful observations to be made about their multifunctionality. the different characteristics that exist at different places within a given With the hierarchical definition in place, structural processors are hierarchy. This, in turn, provides insights into the potential constraints, related to a specific activity which must be defined by an LCI listing. “hot spots” and possible trade-offs that exist when analysing current or Several types of additional data are then also required to enable in­ future systems, energy or otherwise. dicators to be calculated at each processor. The “foreground” scaling The article continues in Section 2 with a description of the use of the information is provided by scenario information such as energy mix and framework, including summarized descriptions of the LCA and MuSIA­ installed capacity data, supplied by outputs from ESMs or other system SEM approaches that form its basis and the various inputs and outputs configuration data. By its nature, this data will differ the most between involved in an ENBIOS simulation. A selection of indicators and possible individual scenarios, where different configurations are being tested, for applications are also discussed. A case study example is then provided in example. Other “background” input data–e.g., employment rates or Section 3, based on a projected “climate neutrality” scenario for the constants for raw material calculations–are likely to remain relatively European energy system using outputs from the Euro-Calliope model. constant from scenario to scenario. The article concludes in Section 4 with a discussion of key outcomes, The key background inputs are taken from LCA databases. Firstly, an potential issues and a roadmap for further development. LCI listing provides a detailed set of information relating to the masses of individual materials, volumes of water and areas of land required as 2. The ENBIOS workflow inputs to a given process–e.g., the production of one unit of energy using a certain technology or process. Outputs to land, water and soil are also While previous efforts have attempted to integrate LCA-based given in relation to radiation, waste and several other aspects. LCI data is assigned at each structural processor in the system dendrogram–one LCI Fig. 1. Overview of general workflow used in the operation of ENBIOS alongside typical inputs and indicators for energy systems. 3 N. Martin et al. Applied Energy 331 (2023) 120366 process per processor–and is provided using the ecospold (.spold) format to produce a set of extensive results; for all structural and functional utilised within the Ecoinvent 3.8 database [28]. processors according to the selected methods as specified in simulation Meanwhile, an LCIA “method” defines the way that LCI listings for a system files. It is important to note here that–unlike most previous at­ given process are transformed into useful final indicators–e.g., global tempts to aggregate LCA and other indicators for complete systems–in warming potential (GWP), total land and water use, and a raft of other ENBIOS we do not perform simple linear aggregations on the results for resource and environmental impact metrics. Many methods exist for individual processors when upscaling to higher hierarchical levels. defining a range of such indicators [29,30]. A number of these LCIA Rather, the input variables themselves are aggregated at each point methods can be used with ENBIOS and the required method must be within the system hierarchy before the calculations are made. That is, chosen prior to initiating calculations. Furthermore, as LCI processes for we aggregate data inputs over the indicators themselves in order to fuel production do not consider the combustion of fuels during their explore emergent relations. In our previous example, extensive in­ final use stage (e.g., the operation of internal combustion engines or dicators are calculated using the applicable LCI data at the structural home heating using natural gas) the additional GHG emissions for fuels processors for wind turbines or solar PV. However, at the functional must also be considered. Here, the required emission factors are taken processor that encompasses these two processors, upscaled LCI data from the Intergovernmental Panel on Climate Change (IPCC) database items would need to be summed before the indicator calculations could [31]. It is noted that similar estimates would need to be added to account proceed. While this will not change indicators derived using linear for the additional contributions from combustion processes when other relationships–e.g., GHG emissions derived using characterization factors air pollution indicators are being used. in LCIA calculations–it is vital for the robustness of the model, and its Aside from life-cycle data, any number of additional socio-metabolic potential suitability to different applications, that separate calculations indicator data sources can also be included, provided it has been nor­ are performed in situations where non-linear relationships are involved. malised for installed capacity or for each unit of energy produced. For A further round of indicators can then be created by relating the example, employment data typically specifies the labour required to initial indicators to additional data about the internal functioning of the maintain a given capacity of electricity and heat infrastructure or pro­ system, thus characterizing the metabolic relationships and constraints duce a certain amount of fuel [32,33]. Indicators can also be calculated that exist within the system. This includes–but is not limited to–the that use raw materials requirements from LCI data in conjunction with derivation of further intensive indicators, which are specific to a given other conversion factors or formulae. Indeed, methods for using LCI data system (e.g., rates, ratios or densities). For example, using the approach to estimate raw material supply risks, end-of-life recycling input rates shown in Fig. 2, one could report “metabolic rate” indicators based on (EoLRIR) and local environmental impact and environmental justice labour requirements or per-unit-of-energy indicators based on the total threats for individual materials have been hypothesised elsewhere amount of energy produced. Indeed, an array of possible intensive in­ [34,35]. In theory, any number of methodologies and sets of input data dicators is possible based on the number of extensive indicators avail­ could be used to create customised indicators for each process within a able. As with the definition of extensive indicators, ENBIOS offers the defined system, and ENBIOS has been specifically formulated to offer user great flexibility to define customised intensive indicators of their high levels of flexibility to users in this regard. Nevertheless, a summary choosing. In this case, we use the most common MuSIASEM indicator of of typical ENBIOS data inputs for energy systems is shown in Fig. 2. system functioning–human activity–using the hours of labour associated with the life cycle of each technology. 2.2. Simulation We calculated all extensive and intensive indicators on a per-unit-of- energy basis, then upscaled and further analysed these indicators Once the system hierarchy has been defined and input parameters examining values at different levels within the system based on the have been specified, indicators can begin to be produced. The first step is defined energy system hierarchy (i.e., the system configuration or Fig. 2. Overview of typical data and methodological inputs and derived final outputs at each processor in an ENBIOS simulation for a given energy system. LCI data, conversion factors, raw material factors and other such data inputs define the system background and are typically entered at the system definition stage and only updated sporadically. Meanwhile, foreground data inputs for individual energy system configurations change according to each scenario being tested. It is noted that input values are aggregated to previous levels of system hierarchies and that calculations always occur at each processor in a system; indicator outputs themselves are never aggregated directly. 4 N. Martin et al. Applied Energy 331 (2023) 120366 “energy mix”). A broad range of indicators can then be examined from 3.2. Data inputs individual or grouped energy sub-technologies to entire energy systems, and vice versa. Results obtained from Euro-Calliope that reflect the information for the system under the “climate neutrality” scenario were used here for 3. Application to European energy scenarios the years 2030 and 2050 [38]. This includes energy production data in terawatt-hours (TWh) and installed capacity data in megawatts (MW) To demonstrate an application of ENBIOS, the framework was for 11 sources of utility-level electricity and three sources of utility-level applied to the European energy system using a set of projected scenario heat. Total production levels (TWh) were also obtained for five sources results for the years 2030 and 2050 obtained from partners within the of fuel supply. Note that no methanol use was included under this SENTINEL project. particular scenario. LCI data was assigned at each structural processor from the Ecoinvent 3.8 database [28]. All electricity processes are 3.1. System definition defined per kilowatt-hour (kWh) of energy produced, while heat pro­ cesses are defined per megajoule (MJ); all inventory items are, thus, The dendrogram for the case study system was defined to align with initially converted to TWh equivalents according to standard conver­ outputs obtained from Euro-Calliope [36], a version of the Calliope sions. Fuel production processes are defined per kilogram (kg), which model [37] being utilised within the project. The model includes all EU requires the data to be converted to energetic equivalents using known member states (except Malta), together with Albania, Bosnia and Her­ MJ/kg calorific value equivalents [39]. As the case study uses total zegovina, Montenegro, Serbia, Switzerland, the United Kingdom, Ice­ energy inputs for the European energy system as a whole, generalised land and Norway. Euro-Calliope simulates all regional processes of LCI process listings for Europe were used, where available. Where these electricity and heat generation at the centralised utility level alongside processes are not available, rest of world (RoW) values are used. How­ the most common forms of fuels used for direct consumption (predom­ ever, in some cases the RoW values deviate significantly from those of inantly those for transport, non-centralised electricity and heat gener­ individual European countries, which are often quite similar. In these ation and use in industrial applications). A representation of the instances, the European country that represents the highest share for dendrogram is shown in Fig. 3. that category in the case study energy mix is used. Lastly, GHG emission Structural processors representing the individual sub-technologies factors for combustion of the five fuels–in kg carbon dioxide equivalent are shown as rounded blocks at the “n-5” level, on the right-hand side (CO2-eq) per kg fuel–were obtained from the IPCC database [31]. A full of the diagram. Functional processors that represent higher level com­ listing of the input data is provided in Table S1 of the supplementary binations of these sub-technologies according to energy supply tech­ information. nology type (“n-4”), renewable status (“n-3”) and energy carrier type Three sample LCIA impact categories from the ReCiPe Midpoint (H) (“n-1”) are then shown as square blocks to the left of this column. group [40] were chosen in this example: GHG emissions in kg CO2-eq were derived using the “GWP100” method, total land occupation in m2 was estimated by summing outputs from the agricultural (“ALOP”) and Fig. 3. ENBIOS dendrogram structure for the EU energy system. Structural processors representing specific electricity, heat and fuel supply processes are located on the right side of the figure. Functional processors are shown on the left of the figure. 5 N. Martin et al. Applied Energy 331 (2023) 120366 urban (“ULOP”) land occupation methods, while water depletion in m3 periods are also included, where appropriate. Consequently, although used the “WDP” method. the lifetimes of energy infrastructure are generally between 20 and 50 Employment data that estimates of the number of full-time jobs years [32], and capacities fluctuate from year to year as equipment is provided by each installed MW of capacity for each electricity and heat implemented and retired, a total number of job positions can be calcu­ generation category [33] was also obtained for each technology. The lated for each moment in time based on current capacities. Data for fuel data includes employment across the manufacturing, construction and production processes is typically given on a per-unit-of-energy basis. installation periods, as well as the ongoing operation and maintenance Hence, the total amount of fuel supplied within a given period–in this tasks occurring within the equipment’s lifetime. Decommissioning case, one year–contributes to the maintenance of a certain number of Fig. 4. Results showing outputs for extensive indicators. Results shown for six indicators across three hierarchical level groupings for projected 2030 and 2050 energy mix outcomes under the EU”climate neutrality” scenario. 6 N. Martin et al. Applied Energy 331 (2023) 120366 positions within that timeframe. A full listing of the utilised data is and emissions created by combusting kerosene (20.4 %) and diesel provided in Table S2 of the supplementary information. (13.8 %) formed from electrolysis processes are also significant. These In order to fully incorporate labour aspects into the metabolic cal­ replace the fossil fuel sources–coal (37.0 %), natural gas (30.1 %), diesel culations within ENBIOS, raw job data was then converted into hours of (13.3 %) and kerosene (10.4 %)–predicted to remain as the dominant human activity (HA) using estimates of annual working hours from emitters in 2030. relevant sectors. Here, “mean weekly hours actually worked” data was Conversely, land occupation increases dramatically, largely driven obtained from the International Labour Organization (ILO) for each by renewable energy sources. This rise is heavily influenced by the use of country represented in the Euro-Calliope model [41]; data is available bioenergy sources, which accounts 71.1 % of the total required land in for all sectors identified within the International Standard Industrial 2050, up from 0.8 % in 2030. Water depletion is expected drop by 38.4 Classification (ISIC) level 2 definitions, of which four are directly %, the changes largely being linked to the move away from fossil fuel applicable to the energy sector and assigned to each processor. Com­ processes with higher water requirements, particularly coal and natural posite annual values for the full model extent–one for each sector–were gas. Meanwhile, total SR more than doubles between 2030 and 2050 in then calculated using the weekly hours worked in individual countries, this example. This is overwhelmingly the result of electricity from wind weighted according to ILO employment rate data. The final data for each and solar sources, which contribute 57.0 % and 39.0 % of the total score sector is listed in Table S3 of the supplementary information. in 2050, respectively. Lastly, the required number of hours of HA from Lastly, raw material factors that enable calculations to be made for employment increases by 78.0 % between 2030 and 2050 under this material supply risk (SR) according to established methodologies scenario. This is, again, largely driven by increases in wind and solar [34,35] were also included. Factors were obtained via external sources installations; wind is the dominant contributor here, rising from a 2.4 % [42,43] for the 55 substances contained within the LCI database that are share of overall activity in 2030 to a 17.1 % share in 2050. considered to be critical raw materials (CRMs) by the European Com­ Additional intensive indicators were then derived by comparing mission (EC) [43]. It should be noted that, although SR values are extensive indicators across and between MuSIASEM hierarchical levels. essentially dimensionless, years (yr) are used as units in accordance with Fig. 5 provides side-by-side comparisons between 2030 and 2050 values the adopted formula. for three “metabolic rate” indicators and four indicators that present extensive attributes on a per-MJ basis. Findings are presented for the three previous levels alongside the total system values at level “n”. A full 3.3. Analysis listing is provided in Table S6 of the supplementary information. A summary of the overall percentage changes at the system level are Results were firstly derived at each of the 19 structural (“n-5”) pro­ shown in Table 2. cessors for a group of six extensive indicators: the total energy produc­ The first of these analyses suggests that the energy metabolic rate tion (directly from Euro-Calliope results), three LCIA indicators, raw (EMR) of the European energy system would drop by 25.6 % between material SR and employment-related human activity. A full listing is 2030 and 2050 under this scenario. This is predominantly the result of provided in Table S4 of the supplementary information. Further calcu­ the large swing to electricity–and wind and solar technologies in par­ lations were then performed in relation to energy supply technology ticular–which rises from 44.6 % to 91.5 % of system energy supply (fuels type (“n-4”), renewable status (“n-3”) and energy carrier type (“n-1”). created via hydrogen from electrolysis are included in electricity totals). The findings are displayed in Fig. 4 and listed in Table S5 of the sup­ And, though heat and fuels are seen to provide significantly higher levels plementary information. A summary of the overall percentage changes of energy per unit of activity, their shares of overall energy production at the system level are shown in Table 1 alongside a listing of the decline from 31.2 % to 7.7 % and 24.2 % to 0.8 %, respectively, during technology type (“n-4”) that makes the most significant contribution to this period. The rapid phasing out of natural gas assumed in the heating overall system change. sector by 2050 results in the sharp drop observed in the EMR value for The extensive data outputs reveal several key findings. The most heat in this period and, coincidentally, a rise in the overall value for immediate trends observed in the energy production data are the overall natural gas as its use as a direct fuel remains relatively high (see Fig. 4). increase of energy production within the system boundary, the move Yet, while more hours of human activity would be required to pro­ towards renewable energy and the increased electrification of the sys­ duce a unit of energy by 2050, findings for the GHG metabolic rate tem by 2050. The breakdown of technologies also reflects the forecast (GHGMR) confirm that significantly less emissions would be produced dominance of electricity from wind (65.5 %) and solar (20.4 %), heat for each of these labour hours. This reduction is again strongly linked to from biomass (7.6 %) and the phasing out of natural gas, which would a substantial change in the share of renewables which, on average, have drop from 40.9 % of system share in 2030 to 0.1 % by 2050. As expected, GHGMR values less than 3 % of those for non-renewables. As a result, net GHG emissions are predicted to drop significantly between 2030 and overall GHGMR values would reduce by 74.6 % between 2030 and 2050. 2050, reducing by 54.7 %; the highest contributor to this drop is coal, Values for the water metabolic rate (WMR)–which decrease at the which drops from 34.8 % of system share in 2030 to zero in 2050. system level by 65.4 % between 2030 and 2050–are, again, strongly The level of emissions produced in 2050 are predominantly linked to influenced by the dramatic drop in the use of fossil fuels. This is espe­ electricity production (43.0 %) and the combustion of fuels produced via cially true of coal and natural gas, whose extensive water use values electrolysis and hydrogen-to-fuel processes (55.8 %); emissions from represented 34.2 % and 15.5 % of the total contribution, respectively, in centralised heat processes are relatively negligible (1.2 %). Solar PV 2030 but are expected to be virtually zero by 2050. A fall in the use of (24.5 %), wind (14.8 %), natural gas (13.3 %) and bioenergy (9. 6 %) nuclear sources of electricity production, from which also have high per- processes are all significant contributors to overall emissions in 2050, Table 1 Summary of changes in extensive indicators between projected 2030 and 2050 scenarios. Overall system changes at the “n” level, in relative percentage, are listed for each indicator. The most significant contributors at the “n-4” level, by change in overall percentage share, are also listed. Total energy GHG emissions Land occupation Water depletion Material supply risk Human activity Change (“n”) +32.5 % − 55.4 % +385.8 % − 38.4 % +146.4 % +78.0 % Most significant Wind Coal Bioenergy Bioenergy Wind Wind contributor Change in share (“n-4”) +58.2 % (7.3 % to − 37.0 % (37.0 % to +70.6 % (0.8 % to +38.8 % (zero to +45.2 % (11.8 % to +14.6 % (2.4 % to 65.5 %) zero) 71.4 %) 38.8 %) 57.0 %) 17.1 %) 7 N. Martin et al. Applied Energy 331 (2023) 120366 Fig. 5. Results showing outputs for intensive indicators. Results shown for seven indicators over complete system and three hierarchical level groupings for projected 2030 and 2050 energy mix outcomes under the EU”climate neutrality” scenario. Table 2 Summary of changes in intensive indicators between projected 2030 and 2050 scenarios. Overall system changes at the “n” level, in relative percentage, are listed for each indicator. Energy metabolic rate GHG metabolic rate Water metabolic rate Water use-to-energy Land use-to-energy GHG-to-energy Supply risk-to-energy Change (“n”) − 25.6 % − 74.9 % − 65.4 % − 53.5 % − 66.3 % +266.7 % +86.0 % 8 N. Martin et al. Applied Energy 331 (2023) 120366 MJ water requirements, is also a factor. Moreover, an increase of tech­ reduction in GHG emissions. However, although water requirements are nologies with relatively low water requirements such as wind and solar unlikely to present serious issues, land occupation, material SR and la­ energy would help to reduce overall water use. Very similar results are bour requirements are all likely to rise dramatically. It is recognised that observed for the water use-to-energy ratio, where the system-wide result the values of land use derived can refer to many different types of use. is seen to reduce by 53.5 % by 2050. For example, the land required by a nuclear plant is vastly different from Meanwhile, the land use-to-energy ratio is predicted to almost land required for a hydropower dam or wind far. Nevertheless, here, the quadruple between 2030 and 2050 under this scenario, predominantly land use totals are largely related to biomass plantations. Meanwhile, due to the dramatic shift towards bioenergy (see Fig. 4). Overall land material issues are strongly linked to wind and solar infrastructure, occupation for the system in 2050 was calculated to be around 4.9 times while the higher labour needs for solar contribute far more than wind or that in 2030, the contribution from bioenergy increasing from 0.8 % in biomass operations. 2030 to 71.4 % in 2050; 51.8 % of this contribution is from heat derived Further analysis of the system, via composite intensive indicators, from biomass. This is also reflected in the dramatic rise in the intensive provided further insights. At the system level, positive outcomes were value of heating witnessed at the “n-1” level. In this sense, it is noted that observed for overall reductions in GHGMR (74.9 %), WMR (65.4 %), natural gas, nuclear energy, diesel and kerosene require considerably water-to-energy (53.5 %) and GHG-to-energy (66.3 %) ratios. Even so, lower areas of land according to this metric. an EMR reduction of 25.6 % suggests that the system would generate less The net amount of GHG emissions produced per unit of supplied energy per unit of human activity in 2050, which could have implica­ energy is a key system indicator for analysing system performance tions on labour markets. Ratios of land use-to-energy and SR-to-energy alongside emissions reduction targets. In this scenario, an overall are both also projected to increase markedly. Indeed, the conse­ reduction of 66.3 % is observed in this indicator between 2030 and quences of different energy transition pathways on both of these issues is 2050. Again, overall system emissions are likely to remain dominated by increasingly being highlighted and could result in wider ecological, non-renewable energy forms in 2030, although by 2050 the dominance political and environmental justice concerns [36,44,45]. of renewables would mean that a relatively large share of emissions Looking specifically to the three key processes at the “n-5” lev­ would also be provided by renewables (51 %). Nevertheless, GHG-to- el–electricity from wind and solar, and heat from biomass–reveals their energy ratios for renewables remain predictably lower than non- influence on overall system indicators. Again, the high labour require­ renewables–around 40 times lower at the “n-3” level in 2050–and wind ment for solar infrastructure has a strong influence on lowering all technologies comfortably provide the best outcomes within the group. metabolic rate values. Very low water requirements for wind turbines However, it is noted that nuclear power is the lowest of the technologies have positive effects on both water-related indicators. Similarly, their examined and produces less than half of the emissions of wind energy. low GHG-to-energy ratios tend to dictate wider outcomes for this indi­ Finally, the level of extensive SR in the 2050 system is almost 2.5 cator. Extremely high land use-to-energy ratios for biomass result in times that of the 2030 system, reflecting a substantial rise in potential their total dominance in this regard. Finally, high SR-to-energy ratios for supply disruptions during this period. Again, these values are rooted wind and solar infrastructure are highly influential on the score re­ predominantly in contributions from wind (57 %) and solar (39 %) ductions for this indicator. technologies. Analysis of SR-to-energy ratios confirms that solar and The case study provides a simple illustration of the ability of ENBIOS wind are substantially higher than all other categories at the “n-4” level; to perform deeper analyses on the different relationships that exist both are between 1.5 and 21.1 times higher than other technologies. within current and future energy systems, relationships that quantify Consequently, a net increase of 86 % is observed in the level of expected constraints and areas of concern across different system levels. While SR per unit of energy for the system as a whole between 2030 and 2050. this example provides a broad demonstration using potential European energy system configurations for illustrative purposes, it is noted that 4. Discussion and conclusions greater detail is possible and that future studies will aim to utilise system dendrograms that incorporate separate consideration of individual re­ The ENBIOS framework brings a systemic approach to the assess­ gions or countries, where LCA data exists. For example, outputs from the ment of environmental impacts and constraints within energy systems Euro-Calliope model are provided for 35 individual countries and using a methodology that combines the high resolution of LCA meth­ separate definitions are often available for LCA processes at the national odologies with the multi-level functionality of the MuSIASEM approach. level. What’s more, shares of energy within technological groups could Furthermore, in the version presented here, we offer a first attempt at be further delineated into sub-technologies where suitable LCA data is systematizing the integration of raw material indicators into energy available. While a lack of data is observed for certain technologies–such modelling practices. Ultimately, ENBIOS has been designed to enable as wind and hydropower–a large selection of different LCA processes are the relationships between indicators at different hierarchical levels to be defined for solar and biomass sub-technologies. This would allow far analysed and the trade-offs between different energy transition path­ more detailed analyses to be undertaken. To that end, an implementa­ ways to be compared–with each other and with defined benchmarks–­ tion and expansion of the ENBIOS framework in a python package has with the aim of informing better energy policy decision making. Full sets been created and is now in beta version [46]. of indicators can be produced for multiple energy system scenarios, In this regard, the validity of the LCIA indicators used in the case derived from different system configurations or across different regions study was tested by comparing them against the range of available re­ and timeframes. Analysis of indicators can derive further information sults for individual European countries. The investigation found that about preferred options, depending on the preferences or perceived most individual values for GHG emissions, land occupation and water limitations of policymakers, and determine whether certain scenarios depletion were within 2–3 % of the values used to represent Europe as a are more–or less–technically feasible than others in terms of land use, whole in the case study. The most significant differences by far were raw material supply issues, employment or other socio-economic fac­ observed for waste incineration, where values could be several times tors. The ability to observe indicator data across and between levels also higher or lower than others. The process used to represent Europe in the allows problem areas such as constraint hotspots to be more easily analysis is for Germany, which is the current and projected largest identified. adopter of waste incineration. As such, the best possible representation The capabilities of ENBIOS were demonstrated using inputs from a is being used. However, it is recognised that a considerable amount of “climate neutrality” scenario for the European energy system in 2030 uncertainty is inherent within the results for waste incineration as a and 2050. Extensive outputs revealed that system changes in this peri­ result of the variability in the regional data. Variations of 10–20 % are od–where a rapid switch is made towards renewables, particularly wind observed in the values for natural gas, coal, wind turbines and solar PV and solar electricity and heat from biomass–would result in a significant cells, suggesting that more regionally detailed investigations are also 9 N. Martin et al. Applied Energy 331 (2023) 120366 likely to improve the accuracy of results for these technologies. result in modelling platforms that include far better representations of It is also noted that in this work we derived sets of indicators using environmental impacts and constraints as we strive to implement system configuration data taken from energy models. However, it is cleaner and more sustainable energy systems as safely, efficiently and hoped that information produced by the framework could also be inte­ rapidly as possible. grated into such models to be used as constrains. This, of course, would require the ENBIOS python module to be included within the broader CRediT authorship contribution statement architecture of an integrated model in order to become a truly interac­ tive element of its system optimisation calculations. In this sense, it is Nick Martin: Conceptualization, Methodology, Formal analysis, also important to note that, while ENBIOS has been primarily formu­ Data curation, Visualization, Writing - original draft, Writing - review & lated to analyse energy systems, it has been designed to be adaptable to editing. Laura Talens-Peiró: Conceptualization, Methodology, Writing any type of hierarchical system and could, theoretically, be used in any – review & editing, Funding acquisition. Gara Villalba-Méndez: number of other applications where multi-level analysis is required, Writing – review & editing, Funding acquisition. Rafael Nebot-Medina: such as agricultural systems. This is especially true if users also require Methodology, Writing - review & editing. Cristina Madrid-López: LCA functionalities to be integrated into their analysis. Conceptualization, Methodology, Visualization, Project administration, In any case, despite the potential of the framework in its current Funding acquisition, Supervision, Writing - review & editing. form, a number of limitations are noted. Firstly, as with many LCA- related applications, assessments are limited to using static informa­ tion based on current inventories. That is, the derived outputs for future Declaration of Competing Interest processes do not contain allowances for future improvements in the background systems that supply energy and material inputs. For The authors declare that they have no known competing financial example, the mix of electricity inputs used in creating or transporting a interests or personal relationships that could have appeared to influence wind turbine in 2040 is essentially “locked” in its current configuration. the work reported in this paper. In reality, many of these inputs would, themselves, produce less emis­ sions or include higher amounts of recycled content as greener energy Data availability practices and circularity initiatives are implemented. It is hoped that the further integration of pLCA concepts–which enable the modification of Data used is in the supplementary tables background systems in modelled environments of this kind–can be included in the python module. This would allow users to manipulate Acknowledgements LCI data assumptions to reflect future developments. Nevertheless, current ENBIOS assessments are capable of providing indications of key The authors would like to thank Prof. Robert U. Ayres for his valu­ bottleneck hotspots in terms of required technological or sourcing im­ able contributions. The research is funded by the EU’s Horizon 2020 provements, using current conditions as reference benchmarks. research and innovation program under the SENTINEL project (GA Similarly, it is acknowledged that variations may well occur to many 837089). Cristina Madrid-López acknowledges the support of the of the input parameters used within the calculations for extensive in­ Spanish Research Agency under the LIVEN project (PID2020-119565RJ- dicators. For example, in the case study presented here, values of SR for I00) individual materials are likely to change over time as reserve amounts and geo-political aspects fluctuate. Likewise, improvements in Appendix A. Supplementary material manufacturing or increased levels of automation may result in lower labour requirements, particularly for newer technologies like wind Detailed listings of key model inputs and case study results are turbines, solar PV cells or biofuel production. Naturally, the extent of available in Tables S1 to S6 of the supplementary information. these improvements is difficult to predict, although learning curves or Supplementary data to this article can be found online at https://doi. other approaches could be applied [47]. However, the architecture of org/10.1016/j.apenergy.2022.120366. ENBIOS means that it is simple to change input parameters of this kind for investigating multiple future scenarios. References Several issues relating to LCA data availability have also been identified. Although Ecoinvent and other major LCI databases contain [1] European Commission. COM(2015) 80 final: Energy union package - several thousand energy-related processes, a lack of good quality data Communication from the Commission to the European Parliament, the Council, the remains for some common processes, especially for newer technologies. European Economic and Social Committee of the Regions and European Investment bank. A framework strategy for a resilient energy union. Brussels: European For example, innovations on wind, solar PV and bioenergy technologies Commission; 2015. are underrepresented and energy storage technologies are not yet rep­ [2] IAMC. Integrated Assessment Modeling Consortium wiki n.d. https://www. resented beyond the production of lithium-ion cells. Accordingly, they iamcdocumentation.eu/index.php/IAMC_wiki. [3] Pang X, Mörtberg U, Brown N. Energy models from a strategic environmental have not been included in the current analysis. Moreover, most quality assessment perspective in an EU context - What is missing concerning renewables? LCA data sources are paywalled, which could seriously restrict access to Renew Sustain Energy Rev 2014;33:353–62. https://doi.org/10.1016/j. calculations. rser.2014.02.005. [4] Giampietro M. Perception and representation of the resource nexus at the interface ENBIOS joins a growing move towards the wider inclusion of sus­ between society and the natural environment. Sustainability (Switzerland) 2018; tainability concepts in energy modelling processes. The key to further 10:2545. https://doi.org/10.3390/su10072545. progress in this area would seem to lie in the ability to place related data [5] Süsser D, Martin N, Stavrakas V, Gaschnig H, Talens-Peiró L, Flamos A, et al. On the relevance of environmental and social factors in energy modelling: Variables to and operations into environments that are also compatible with the include and effects of omitting them. Energy Res Soc Sci 2022; Manuscript. models themselves. As such, open-source applications such as Wurst [6] Süsser D, Gaschnig H, Ceglarz A, Stavrakas V, Flamos A, Lilliestam J. Better suited [48], PREMISE [19] and the ENBIOS module that operate within Python or just more complex? On the fit between user needs and modeller-driven environments are making it easier to import model output data and improvements of energy system models. Energy 2022;239:121909. https://doi. org/10.1016/j.energy.2021.121909. manipulate and automate the processing of life cycle data. Again, the [7] Finkbeiner M, Inaba A, Tan RBH, Christiansen K, Klüppel H-J. The new ability to return outputs from applications such as ENBIOS back into international standards for life cycle assessment: ISO 14040 and ISO 14044. Int J energy models directly to achieve genuine two-way synthesis would Life Cycle Assess 2006;11:80–5. https://doi.org/10.1065/lca2006.02.002. [8] Arvesen A, Luderer G, Pehl M, Bodirsky BL, Hertwich EG. Deriving life cycle greatly improve the ability of models to integrate the power of LCA and assessment coefficients for application in integrated assessment modelling. Environ other high resolution environmental data in this manner. This would Model Softw 2018;99:111–25. https://doi.org/10.1016/j.envsoft.2017.09.010. 10 N. Martin et al. Applied Energy 331 (2023) 120366 [9] Laurent A, Espinosa N, Hauschild MZ. LCA of energy systems. In: Hauschild MZ, [27] Giampietro M, Mayumi K. Multiple-scale integrated assessment of societal Rosenbaum RK, Olsen SI, editors. Life cycle assessment: Theory and practice. metabolism: Introducing the approach. Popul Environ 2000;22:109–53. https:// Cham: Springer; 2018. p. 633–68. https://doi.org/10.1007/978-3-319-56475-3_ doi.org/10.1023/A:1026691623300. 26. [28] Ecoinvent. Ecoinvent version 3.8 (2021) database 2021. https://v38.ecoquery. [10] Valente A, Iribarren D, Dufour J. Comparative life cycle sustainability assessment ecoinvent.org/Home/Index. of renewable and conventional hydrogen. Sci Total Environ 2021;756:144132. [29] Jungbluth N. Description of life cycle impact assessment methods: Supplementary https://doi.org/10.1016/j.scitotenv.2020.144132. information for tenders. Schaffhausen: ESU-services Ltd.; 2021. [11] Junne T, Simon S, Buchgeister J, Saiger M, Baumann M, Haase M, et al. [30] Rosenbaum RK. Overview of existing LCIA methods - Annex to Chapter 10. In: Environmental sustainability assessment of multi-sectoral energy transformation Hauschild MZ, Rosenbaum RK, Olsen SI, editors. Life cycle assessment: Theory and pathways: Methodological approach and case study for Germany. Sustainability practice. Cham: Springer; 2018. p. 1147–83. https://doi.org/10.1007/978-3-319- (Switzerland) 2020;12:1–28. https://doi.org/10.3390/su12198225. 56475-3_40. [12] Luderer G, Pehl M, Arvesen A, Gibon T, Bodirsky BL, de Boer HS, et al. [31] IPCC. Emission factor database 2021. https://www.ipcc-nggip.iges.or.jp/EFDB/ Environmental co-benefits and adverse side-effects of alternative power sector main.php (accessed April 21, 2021). decarbonization strategies. Nat Commun 2019;10:1–13. https://doi.org/10.1038/ [32] Ram M, Aghahosseini A, Breyer C. Job creation during the global energy transition s41467-019-13067-8. towards 100% renewable power system by 2050. Technol Forecast Soc Change [13] Reinert C, Schellhas L, Mannhardt J, Shu DY, Kämper A, Baumgärtner N, et al. 2020;151:119682. https://doi.org/10.1016/j.techfore.2019.06.008. SecMOD: An open-source Modular framework combining multi-sector system [33] Rutovitz J, Dominish E, Downes J. Calculating global energy sector jobs: 2015 optimization and life-cycle assessment. Front Energy Res 2022;10:884525. https:// methodology update. Sydney: Institute for Sustainable Futures, University of doi.org/10.3389/fenrg.2022.884525. Technology Sydney; 2015. [14] Tokimatsu K, Tang L, Yasuoka R, Ii R, Itsubo N, Nishio M. Toward more [34] Martin N, Madrid-López C, Villalba-Méndez G, Talens-Peiró L. New techniques for comprehensive environmental impact assessments: Interlinked global models of assessing critical raw material aspects in energy and other technologies n.d.;Under LCIA and IAM applicable to this century. Int J Life Cycle Assess 2020;25:1710–36. revi. https://doi.org/10.1007/s11367-020-01750-8. [35] Talens-Peiró L, Martin N, Villalba-Méndez G, Madrid-López C. Integration of raw [15] Arvidsson R, Tillman AM, Sandén BA, Janssen M, Nordelöf A, Kushnir D, et al. materials indicators of energy technologies into energy system models. Appl Environmental assessment of emerging technologies: Recommendations for Energy 2022;307:118150. https://doi.org/10.1016/j.apenergy.2021.118150. prospective LCA. J Ind Ecol 2018;22:1286–94. https://doi.org/10.1111/ [36] Tröndle T, Lilliestam J, Marelli S, Pfenninger S. Trade-offs between geographic jiec.12690. scale, cost, and infrastructure requirements for fully renewable electricity in [16] Gibon T, Wood R, Arvesen A, Bergesen JD, Suh S, Hertwich EG. A methodology for Europe. Joule 2020;4:1929–48. https://doi.org/10.1016/j.joule.2020.07.018. integrated, multiregional life cycle assessment scenarios under large-scale [37] Pfenninger S, Pickering B. Calliope: A multi-scale energy systems modelling technological change. Environ Sci Technol 2015;49:11218–26. https://doi.org/ framework. J Open Source Softw 2018;3:825. 10.1021/acs.est.5b01558. [38] Pickering B, Lombardi F, Pfenninger S. Diversity of options to reach carbon- [17] Hertwich EG, Gibon T, Bouman EA, Arvesen A, Suh S, Heath GA, et al. Integrated neutrality across the entire European energy system. Joule 2022;6:1253–76. life-cycle assessment of electricity-supply scenarios confirms global environmental https://doi.org/10.1016/j.joule.2022.05.009. benefit of low-carbon technologies. Proc Natl Acad Sci U S A 2015;112:6277–82. [39] Eurostat. Energy data: 2020 edition. Luxembourg: Publications Office of the https://doi.org/10.1073/pnas.1312753111. European Union; 2020. https://doi.org/10.2785/68334. [18] Pehl M, Arvesen A, Humpenöder F, Popp A, Hertwich EG, Luderer G. [40] Huijbregts MAJ, Steinmann ZJN, Elshout PMF, Stam G, Verones F, Vieira M, et al. Understanding future emissions from low-carbon power systems by integration of ReCiPe2016: A harmonised life cycle impact assessment method at midpoint and life-cycle assessment and integrated energy modelling. Nat Energy 2017;2:939–45. endpoint level. Int J Life Cycle Assess 2017;22:138–47. https://doi.org/10.1007/ https://doi.org/10.1038/s41560-017-0032-9. s11367-016-1246-y. [19] Sacchi R, Terlouw T, Siala K, Bauer C, Cox B, Daioglou V, et al. PRospective [41] International Labour Organization. ILOSTAT data catalogue 2022. https://ilostat. EnvironMental Impact asSEment (premise): A streamlined approach to producing ilo.org/data/. databases for prospective life cycle assessment using integrated assessment models. [42] Wendling ZA, Emerson JW, de Sherbinin A, Esty DC. Environmental performance Renew Sustain Energy Rev 2022;160:112311. https://doi.org/10.1016/j. index 2020. New Haven, CT: Yale Center for Environmental Law & Policy; 2020. rser.2022.112311. [43] European Commission. Study on the EU’s list of critical raw materials (2020) - [20] Dirnaichner A, Rottoli M, Sacchi R, Rauner S, Cox B, Mutel C, et al. Life-cycle Final report. Luxembourg: Publications Office of the European Union; 2020. impacts from different decarbonization pathways for the European car fleet. https://doi.org/10.2873/11619. Environ Res Lett 2022;17:044009. https://doi.org/10.1088/1748-9326/ac4fdb. [44] Lèbre É, Stringer M, Svobodova K, Owen JR, Kemp D, Côte C, et al. The social and [21] di Felice LJ, Ripa M, Giampietro M. An alternative to market-oriented energy environmental complexities of extracting energy transition metals. Nat Commun models: Nexus patterns across hierarchical levels. Energy Policy 2019;126:431–43. 2020;11:1–8. https://doi.org/10.1038/s41467-020-18661-9. https://doi.org/10.1016/j.enpol.2018.11.002. [45] Bobba S, Carrara S, Huisman J, Mathieux F, Pavel C. Critical raw materials for [22] Parra R, di Felice LJ, Giampietro M, Ramos-Martin J. The metabolism of oil strategic technologies and sectors in the EU: A foresight study. Luxembourg: extraction: A bottom-up approach applied to the case of Ecuador. Energy Policy Publications Office of the European Union; 2020. https://doi.org/10.2873/ 2018;122:63–74. https://doi.org/10.1016/j.enpol.2018.07.017. 865242. [23] SENTINEL. Sustainable Energy Transitions Laboratory (SENTINEL) n.d. https://se [46] Nebot-Medina R, Martin N, Talens-Peiró L, Madrid-López C. The ENvironmental ntinel.energy/. and BIOeconomic System Assessment (ENBIOS) module n.d. [24] Giampietro M, Mayumi K, Ramos-Martin J. Multi-scale integrated analysis of [47] Martin N, Madrid-López C, Villalba G, Talens-Peiró L. Overlooked factors in societal and ecosystem metabolism (MuSIASEM): Theoretical concepts and basic predicting the transition to clean electricity. Environ Res: Infrastruct Sustain 2022; rationale. Energy 2009;34:313–22. https://doi.org/10.1016/j. 2:021005. https://doi.org/10.1088/2634-4505/ac70f7. energy.2008.07.020. [48] Mendoza Beltran A, Cox B, Mutel C, van Vuuren DP, Font Vivanco D, Deetman S, [25] Giampietro M, Mayumi K. A dynamic model of socioeconomic systems based on et al. When the background matters: Using scenarios from integrated assessment hierarchy theory and its application to sustainability. Struct Chang Econ Dyn 1997; models in prospective life cycle assessment. J Ind Ecol 2020;24:64–79. https://doi. 8:453–69. https://doi.org/10.1016/S0954-349X(97)00017-9. org/10.1111/jiec.12825. [26] Giampietro M, Mayumi K. Multiple-scale integrated assessment of societal metabolism: Integrating biophysical and economic representations across scales. Popul Environ 2000;22:155–210. https://doi.org/10.1023/A:1026643707370. 11