Energy Policy 126 (2019) 431–443 Contents lists available at ScienceDirect Energy Policy journal homepage: www.elsevier.com/locate/enpol An alternative to market-oriented energy models: Nexus patterns across T hierarchical levels Louisa Jane Di Felicea, , Maddalena Ripaa, Mario Giampietroa,b ⁎ a Institute of Environmental Science and Technology, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain b ICREA, Pg. Lluís Companys 23, 08010 Barcelona, Spain ARTICLE INFO ABSTRACT Keywords: From a biophysical perspective, energy is central to the behaviour of social-ecological systems. Its ubiquity MuSIASEM means that energy is entangled with nexus elements, including water, land, emissions and labour. At the science- Societal Metabolism policy interface, large market-oriented energy models dominate as the tool to inform decision-making. The Water-energy-food nexus outputs of these models are used to shape policies, but strongly depend on sets of assumptions that are not Hierarchy Theory available for deliberation and gloss over uncertainties. Taking an approach from complexity, we propose an Science-Policy Interface Complexity alternative to market-oriented energy models, describing the behaviour of energy systems in relation to patterns of nexus elements across hierarchical levels. Three characteristics are central to the approach: (i) the distinction of the model's building blocks into functional and structural elements; (ii) their hierarchical organisation and (iii) the description of nexus patterns at each level, through the tool of the processor. To illustrate the model, it is applied to Catalonia's energy sector, linking production and consumption patterns. The framework may help inform stakeholder deliberation on pressing energy and nexus issues. 1. Introduction assessments that handle multiple types of variables at once, such as water, climate, land, food and economic variables. These kinds of ap- A biophysical reading of societies describes their behaviour in re- proaches can be grouped under the umbrella term of nexus assessments. lation to the extraction, exchange, distribution and consumption of The surge in popularity of the nexus in academic discourses has been material flows (Cleveland, 1987). From a biophysical perspective, en- rapid (Bazilian et al., 2011; Ringler et al., 2013; Howells et al., 2013; ergy lies at the core of how societies function, providing the input for all Endo et al., 2015; Howarth and Monasterolo, 2016; Kurian, 2017), and economic processes and regulating the interface between societies and fits within scientific moves towards holistic approaches and inter- natural ecosystems (Ostwald, 1907; Lotka, 1922; White, 1943; Cottrell, disciplinarity (Lam et al., 2014). In relation to nexus assessments, 1955; Odum, 1971; Georgescu-Roegen, 1971; Smil, 2010). The ubiquity Stirling (2015) highlights the need to view sustainability challenges as of energy means that it is closely linked to pressing problems faced by complex ones, where “it is ever more clearly understood that circum- social-ecological systems (SESs), including climate change, security, scribed monodisciplinary or single-sector approaches are not enough”. resource depletion, justice and poverty. At the science-policy interface, and in particular in EU policy, Two recent trends in energy studies and assessments have aimed at moving away from the monodisciplinary, mostly technical, arena to moving the framing of energy issues from a single-sector, mono- which energy has been historically confined to has not been simple disciplinary one, towards integrated and holistic descriptions. The first (Giampietro et al., 2013). When it comes to framing energy beyond its trend, at the level of energy studies, is a move from the dominant en- biophysical dimension and considering it as a social issue, some pro- gineering framing of energy to one that includes wider social views gress has been made with the recent Clean Energy for All Europeans (Sovacool, 2014), recognising the links of energy to social issues and to package (European Commission, 2016a), which includes discourses of the functioning of society beyond its biophysical dimension. The second energy poverty and a focus on citizens’ rights with respect to energy trend, at the level of energy assessments, is a move towards integrated services. When it comes to handling multiple elements linked to energy Abbreviations: SES, Social-Ecological System; GHG, Greenhouse gases; PES, Primary Energy Source; EC, Energy Carrier; EU, End Use; MuSIASEM, Multi-Scale Integrated Analysis of Societal and Ecosystem Metabolism; CUF, Capacity Utilization Factor; OECD, Organisation for Economic Co-operation and Development; WELMM, Water, energy, land, materials and manpower; PWR, Pressurised water reactor; PV, Photovoltaic; CC, Combined cycle; CHP, Combined heat and power; WWTR, Waste water treatment and reuse; PHS, Pumped hydroelectric storage ⁎ Corresponding author. https://doi.org/10.1016/j.enpol.2018.11.002 Received 20 February 2018; Received in revised form 24 September 2018; Accepted 2 November 2018 Available online 04 December 2018 0301-4215/ © 2018 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/BY/4.0/). L.J. Di Felice et al. Energy Policy 126 (2019) 431–443 through nexus assessments, a linearly interactive (interdisciplinary) ap- notion of holons (Koestler, 1968, 1978), and between funds and flows, proach has been taken to inform EU policy, whereby different models following Georgescu-Roegen's flow-fund model (Georgescu-Roegen, specialising in the accounting of different elements provide the input 1971). We explain the model through the case study of Catalonia's for other models specialising in other elements, and so on. The EU 2016 energy sector. By introducing numbers associated to a case study, we Reference Scenario, the EU's most recent modelling exercise informing show how the methodology could be suited in addressing different policy-makers on energy, transport and climate issues, gives an example types of policy questions. of the concatenation of inputs and outputs between models: “PRIMES The rest of this paper is structured as follows: Section 2 provides a uses as inputs macroeconomic and multi-sectorial projections from review of current energy assessment frameworks dealing with multiple GEM-E3 and projections of world energy prices from PROMETHEUS. nexus elements (2.1) and introduces the key principles of energy ana- PRIMES conveys projections to GAINS, GEM-E3 and CAPRI” (European lysis in MuSIASEM (2.2); Section 3 outlines the rationale of the model, Commission, 2016b, p.16). PRIMES is a market-oriented energy model, starting with its building blocks (3.1) and explaining how the building GAINS is a GHG emission model, GEM-E3 is a general equilibrium blocks are linked across hierarchical levels (3.2), through which macroeconomic model and CAPRI is an agricultural sector model.1 equation (3.3) and within which boundaries (3.4). Section 4 introduces Whilst concatenating inputs and outputs of different models pro- the data used in the example (4.1) and provides an overview of the case vides a form of interdisciplinarity, the inner workings of the models study (4.2). Results and Discussion are presented in Section 5, starting themselves are not open for deliberation. Taking a view from com- with the production mix (5.1) and linking it to consumption patterns plexity, which can be defined as the existence of plural, legitimate and across different economic sectors (5.2). The discussion section focuses irreducible representations of the same issue (Simon, 1962; Whyte on the model's uncertainty (5.3) and on its policy relevance (5.4). et al., 1969; Rosen, 1991, 2000; Ahl and Allen, 1996; Kovacic, 2017), we argue that moves towards interdisciplinarity in energy studies 2. Background and literature review should recognise that the same energy issue can be viewed with dif- ferent framings depending on the stakeholder's perspective and prio- 2.1. Energy-nexus assessments rities. As a consequence, models informing decision-making should allow for deliberation at various steps of the analysis, rather than From a systems perspective, various modelling tools integrating providing a normative set of outputs. According to complex systems energy assessments with other nexus elements have been proposed, as theory, different quantitative representations depend on different pre- summarised by Keairns et al. (2016). These include the Climate Land- analytical choices of scale and dimensions of analysis – i.e. the adoption use Energy and Water Strategy (CLEWS) modelling framework, the of non-equivalent descriptive domains (Giampietro et al., 2006). This Stockholm Resilience Center's Water Energy and Planning (WEAP) implies that the results of quantitative analyses developed within non- system, the Institute for Applied System Analysis’ Model for Energy equivalent descriptive domains are not reducible to each other (Rosen, Supply Systems and their General Environmental Impact (MESSAGE), 1991). This conceptual impossibility forces analysts to adopt a series of the Integrated Model to Assess the Global Environment (IMAGE), the “heroic assumptions” about expected relations associated with large Water Energy Land Material and Manpower (WELMM) and, finally, the doses of uncertainty (Funtowicz and Ravetz, 1990; Pereira & Multi-Scale Integrated Analysis of Society and Ecosystem Metabolism Funtowicz, 2015). In the models informing EU policy, assessment of the (MuSIASEM). Of these, CLEWS, MESSAGE and IMAGE are integrated robustness of their assumptions and implications are impossible as the assessment models (IAMs) with common characteristics: they produce modelling codes are not available to the public, such as the PRIMES long term scenarios with a global focus, employ optimisation models, modelling codes used to build EU energy scenarios. Hidden assump- are highly detailed and, although describing the interactions between tions, which could become contested if made explicit, lead market-or- human and natural ecosystems, they do not consider labour as a vari- iented energy models to “provide normative optimised scenarios, in able. IAMs are widely used to develop scenarios for the IPCC as they which real implementation bottlenecks are ignored (e.g. uncertainty, model climate change in relation to a diverse set of social and physical heterogeneity of decision makers and market imperfections)’’ (Dodds variables. WEAP, on the other hand, is mostly concerned with in- et al., 2015, p. 85). This is a general problem associated with the tegrated assessments of water. concept of so-called evidence-based policy (Ravetz, 1971; Rayner, 2012; WELMM and MuSIASEM present many similarities among them, Saltelli and Giampietro, 2017). and differences to the IAMs mentioned above. Their common char- Thus, our model aims to assess nexus elements simultaneously acteristics are the disaggregation of types of energy (electricity, heat within a coherent framework capable of integrating information and fuels), which are considered as separate variables, the introduction coming from different descriptive domains, and from different stake- of labour as a variable in the system and a simple analytical approach holders. In this way, we avoid what Cairns and Krzywoszynska (2016) that is accessible to decision-makers. Operating at a scale of middle have defined as integrated imaginaries, whereby nexus assessments are numbers (between the process scale and the system level), MuSIASEM used to push for technocratic win-win solutions. All the opposite, a distinguishes itself for the lack of closed semantic descriptions of its transparent framing of assumptions across hierarchical levels may categories, which may be decided with stakeholders, and for using a generate uncomfortable knowledge for policymakers, showing that coarse-grained approach that allows understanding the big picture (at a win-win solutions are rarely possible. given level of representation) without losing details that are relevant to The aim of the paper is methodological: building on the Multi-Scale the policy process (at a different level of representation). In the next Integrated Analysis of Societal and Ecosystem Metabolism (MuSIASEM) section, the main characteristics of energy assessments in MuSIASEM, (Giampietro et al., 2012, 2013, 2014), we propose a method of in- on which our model builds on, are outlined. tegration of quantitative information that handles multiple nexus ele- ments across different levels of the energy system. What defines the 2.2. MuSIASEM model is not the choice of categories and variables, but the way in which nexus patterns are linked across hierarchical levels. Two key The foundation of energy assessments with MuSIASEM is to ac- distinctions are maintained throughout the analysis: between structural knowledge the complexity of the concept of energy, which is semantic and functional elements of the energy system, following the Koestler's in its nature. MuSIASEM calls for at least three non-equivalent ways of accounting for energy flows consumed by society, all of which are needed to assess the performance of an economy (Giampietro et al., 1 For an overview of the main characteristics of each model, see (European 2013). Primary Energy Sources (PES) are what is required from Commission, 2018). boundary conditions - they are sources outside of human control, such 432 L.J. Di Felice et al. Energy Policy 126 (2019) 431–443 as the amount of oil and coal in the earth's lithosphere; energy Carriers (EC) are the flows used for transformations under human control, for example electricity, heat and fuels; Energy end-uses (EUs), finally, are characterisations of a specific pattern of energy carriers for a specific task (what the energy carriers are used for in society). In MuSIASEM this characterisation requires combining assessments of quantities of energy to assessments of quantities of other inputs required for achieving an expected task. In addition to the conceptualisation of different energy flows describing energy transformations in society, MuSIASEM distinguishes between two types of energy forms: thermal energy (e.g., MJ of fuels) and mechanical energy (e.g., kWh of elec- tricity). This distinction applies to both primary energy sources and energy carriers. Finally, MuSIASEM categorises relevant elements of energy systems into funds, flows and stocks, following Georgescu- Roegen's flow-fund model. Funds are elements whose identity remains intact over the chosen scale of analysis, flows are elements that either enter the system without exiting or exit it without entering, and stocks Fig. 1. Examples of a structural and a functional processor - Nuclear PWR and are non-renewable supplies from which flows can be extracted. On a Baseload electricity generation. yearly timescale of the energy system, for example, labour and land are funds, water and electricity are flows, and oil reserves are stocks. Internal inputs are elements that are consumed or produced (flows) and Recently, MuSIASEM has evolved to include the tool of the pro- maintained (funds) by society, therefore under human control – they cessor, i.e. an element in the metabolic network that processes a series are inside the technosphere. They are represented on the upper half of of inputs and generates a series of outputs at a given hierarchical level. the processor: flow elements on the left and fund elements on the right. The concept of processor was first introduced in relational analysis by External flows are elements that are produced or received (flows) by Robert Rosen as a way to identify the relation between functional and processes outside human control – they are metabolised outside the structural elements, arising from the definition of a shared final cause technosphere, requiring a supply and sink capacity (funds or stocks) to (purpose) (Louie, 2010). In this sense, processors represent a bridge be provided by the biosphere. They are represented on the lower half of between semantic and quantitative definitions. They have been im- the processor: flows associated with a required supply capacity on the plemented within the MuSIASEM framework and applied to the de- left, and flows associated with a required sink capacity on the right. scription of Brazil's oil and gas sector by Aragão and Giampietro (2016), There is no reference to fund elements on the lower half of the pro- by González-López and Giampietro (2017) to the case of charcoal cessor, as the fund elements required for guaranteeing the supply ca- production in Mexico, by González-López and Giampietro (2018) to the pacity and sink capacity belong to the biosphere (they are operating in case of the gas and oil sector in Mexico and by Parra et al. (2018) to the processes outside human control). description of Ecuador's oil extraction metabolism. At a larger scale, Each processor built for the analysis is associated with five sets of processors have been used to anticipate changes in the social structure elements: in relation to agriculture by Giampietro (2018). While their framing as a tool for nexus analysis is novel, similar tools have been used under 1. Flows under human control, coming to the processor from the different names in other fields (see, for example, nodes in neural net- technosphere, for example electricity from the grid consumed in the works, enzymes in biochemistry, and production functions in nuclear plant; economy). A notable example of a previous conceptualisation of a 2. Funds under human control, embedded in the technosphere, such as processor is the “resource processing system” by Grenon (1978) as a the land on which the power plant is built; tool for the WELMM method. However, to the authors’ best knowledge, 3. Flows extracted from the biosphere, such as the freshwater ab- Grenon's tool has not been applied in recent times to carry out nexus stracted for cooling; assessments. 4. Flows released into the biosphere, e.g. GHG emissions; 5. Flows generated by the processor for societal consumption, in this 3. Methodology example a flow of baseload electricity. 3.1. Building blocks of the model: structural and functional processors Flows are either produced by funds or extracted from stocks, therefore they relate funds and stocks to processes. Funds, on the other Building on MuSIASEM's recent advancements, our model's building hand, are shared. Shared funds in the biosphere have a level of sink blocks are processors of the energy system, split into functional and capacity that must be maintained, while funds in the technosphere are structural ones. A functional processor can be defined in notional terms shared among the different processes taking place in society. This im- as a node in a network that metabolises inputs from the technosphere plies the existence of an internal constraint – i.e. an opportunity cost - and the biosphere, and produces outputs, both useful (to society) and for the utilisation of internal inputs in the processors: the amount of released to the environment (e.g. emissions). At the same time, we can labour or land that a process can use depends on how much is being describe a structural processor, in technical terms, as the profile of inputs used by others. For example, limits to the amount of land available for and outputs associated with the operation of a structural element – i.e. a biofuel crops can be defined by the amount of land needed for agri- given technology expressing a biophysical set of transformations. Fig. 1 culture.2 External constraints to the behaviour of the metabolic pattern shows an example of two of the processors of this case study, the structural processor of a nuclear pressurised water reactor (PWR) and the functional processor of generation of baseload electricity. There is 2 While the limit of total land available in a country is an external one, outside no 1:1 mapping between the characteristics of processors describing of human control, limits as to how much land is available for a given process structures and functions: the same function can be covered by a mix of given the land used for another are internal ones, since they are not dictated by different structures, and the same structure can cover more than one the biosphere but by the allocation of shared funds across different sectors – function. there may be enough land to produce biofuels (no external limits), but it may The processor relates internal inputs and outputs to external ones. already be occupied by agricultural production (internal limit). 433 L.J. Di Felice et al. Energy Policy 126 (2019) 431–443 Fig. 2. Example of a hierarchical organisation of the energy system. of societies are determined by limits in the biosphere, such as the ex- the same function can be produced by a different combination of haustion of the stock of crude oil on the supply side, or the excessive structures. amount of GHGs released into the atmosphere on the sink side. 3.3. Modelling equation 3.2. Relations between processors across hierarchical levels Data taken at the structural level is aggregated and combined to describe the behaviour of the energy system at the functional level, and A hierarchy is “a system that is composed of interrelated sub- at the whole, moving from the lower level to the upper level in the systems, each of the latter being, in turn, hierarchic in structure until hierarchical map. The relations among different hierarchical levels we reach some lowest level of elementary subsystem” (Simon, 1962, p. follow simple rules of aggregation. The equation below shows how 468). Fig. 2 illustrates the energy system broken down across hier- patterns of nexus elements are scaled across levels of analysis. archical levels, populated by structural and functional processors. In this case the hierarchical organisation is nested, as all the lower-level Ei, n = Ei, n 1 Sn 1, n Pi components are aggregated into higher level ones, and organisational, i as moving up in hierarchical levels does not necessarily imply moving Ei,n represents the set of nexus elements at the level n, so could be, up in size, and aggregations between elements are purely definitional for example, water use or GHG emissions, depending on the index i. The and context-dependent. value at a given hierarchical level n is equal to the aggregation of all Connecting components of the hierarchy, hierarchical pathways are values of the same nexus elements at the level n-1, multiplied by a vertical ones that move up or down levels, linking primary energy scaling factor Sn-1,n and the relative contribution of each element, ex- sources (uranium and coal, in the figure), to technologies producing pressed as a percentage, Pi. This equation holds for intensive variables, and consuming energy carriers and to their purposes. Therefore, they for example GHG emissions expressed in terms of emissions per kWh of cross the border between the biosphere and the technosphere. electricity produced. An intensive description of the sytem's processes is Sequential pathways, then, are those operating in sequence at the same necessary to build scenarios. When the model is used for descriptive level to fulfil a unitary function. For example, in order to provide purposes, processes may be described in extensive terms. When ag- electricity to society, resources must be extracted, then converted to gregating extensive variables across hierarchical levels, a simple sum of electricity, and finally transmitted and distributed. the lower level elements is required (therefore both S and P are set to In the figure, only one hierarchical pathway has been expanded, one). The scaling factor takes into account economy of scale effects from uranium and coal, to nuclear and coal plants, to the generation of when moving up in size, and is needed when using the model to build baseload electricity3 needed in society. scenarios. Given the descriptive purpose of the case study of Catalonia, Functional processors are represented in Fig. 2 as rectangular, while the scaling factor has been kept to one across levels. structural ones are oval. By structure, we refer to all parts of the system that have been realised in a physical instance. For example, specific technologies or human beings are structures. Functional elements are 3.4. Boundaries of the case study aggregations of structural ones based on the role they play in the system. Both nuclear plants and coal plants can be used to provide The modelling framework does not imply a specific set of bound- baseload electricity to society, therefore the two structural elements can aries, which can be expanded or compressed depending on the goal of be grouped into a functional one occupying a higher hierarchical level the analysis. The case used as an example here is the energy sector of and labelled “generation of baseload electricity”. In this organisation, the Autonomous Community of Catalonia, located on the north-eastern as mentioned previously, there is no 1:1 mapping between structures extremity of the Iberian Peninsula, for the year 2012. For the chosen and functions: the same structure can be used to cover multiple func- case study, we do not consider every step of the energy system's se- tions, such as refineries producing heating fuels and transport fuels, and quential pathways, but focus on describing the steps that happen within the geographical boundaries of Catalonia, i.e. production of electricity, heat and fuels. There is some minor extraction of crude oil in Catalonia 3 See footnote 4. that is not considered in this instance. To further simplify our example, 434 L.J. Di Felice et al. Energy Policy 126 (2019) 431–443 we do not describe transport, transmission and distribution of energy Table 1 carriers, focusing on the production and consumption of energy carriers Top-down characterisation of Catalonia's energy system. within the geographical boundaries of Catalonia. Source: ICAEN (2012), Focal level. Type Label Unit Production Imports- Exports 4. Case study Primary Energy Source Coal ktoe 23 10 4.1. Data sources Primary Energy Source Oil ktoe 149 10113 Primary Energy Source Natural Gas ktoe 1 5961 Primary Energy Source Non-renewable ktoe 112 0 To frame the description of Catalonia's energy sector, in addition to industrial waste bottom-up data, top-down data is also collected. We refer to top-down Primary Energy Source Uraniuma tU 0 63 data as data collected through statistical bodies, and to bottom-up data Energy carrier Refinery gases ktoe 0 0 as data collected for instances of power plants, refineries and other Energy carrier GLP ktoe 364 260 structures, mostly through reports. Top-down energy statistics were Energy carrier Naphthab ktoe -27 2157 Energy carrier Gasolines ktoe 1362 -495 collected from the Catalan Institute of Energy (ICAEN). Different Energy carrier Kerosene ktoe 871 244 sources of bottom-up data were used to characterise processors.4 When Energy carrier Gas-oil ktoe 2991 840 possible, data specific to individual power plants, and of the correct Energy carrier Fuel-oil ktoe 1588 -1193 year (2012) were used. Neither data collected through reports nor data Energy carrier Biogas ktoe 64 0 Energy carrier Biofuels ktoe 28 303 obtained through statistical offices include uncertainty ranges, and Energy carrier Electricity TWh 45 4 since the focus of the case study is to explain the methodology rather than to present numbers, a sensitivity analysis of the numerical inputs a Imports of uranium are not provided in statistical tables and have been and outputs is beyond our scope. However, Section 5.3 presents a derived by assuming that 0.73 kg of Uranium are needed for each TJ produced methodological discussion of uncertainty within the model. in a nuclear pressurised water reactor (PWR) (Dones et al., 2007). b Negative values in the production of energy carriers refer to fuels and heat products that are consumed in the intermediate refinery process, where the 4.2. Overview of Catalonia's energy system total consumption is higher than the refinery output (since only net outputs are available through ICAEN, similarly to how only net imports are available, Table 1 shows a summary of top-down statistics characterising leading to negative imports-exports balances). Catalonia's energy sector in 2012. In the first part of the table, extrac- tion data of PES, as well as net imports, are provided. Following the MuSIASEM distinction between PES and EC, the second part shows Table 2 refinery process outputs of fuels and heat products, and production of Characterisation of Catalonia's electricity sector. electricity, with their respective net imports. Catalonia produces most Source: ICAEN (2012) Level −1. of its electricity, while importing almost all of its crude oil and natural # of Installed Electricity CUFa gas supply. For heating and transport fuels, there are differences among plants Capacity (GWh) (%) different elements, with a high production of gasolines and gas-oils and (MW) imports of naphtha, fuel-oils and biofuels. This characterisation is si- Natural gas combined cycle (CC) 9 4112 8342 23 milar to other EU countries that are poor in natural resources and have Nuclear 3 3147 23,996 87 a large amount of power plants as well as complex refining capacity,5 Hydropower 335 2361 3653 18 therefore importing PES to produce EC. Wind 44 1258 2691 24 To have on overview of the process of electricity production within Combined Heat & Power (CHP) 135 1021 5896 66 Solar PV N/A 249 406 19 the boundaries of Catalonia, Table 2 is introduced. Each type of power Pig Manure 6 92 738 92 plant is associated with its installed capacity, gross electricity genera- WWTR 16 66 413 72 tion and capacity utilisation factor (CUF).6 Nuclear power produces Urban Solid Waste 4 46 139 34 over 50% of Catalonia's electricity, and CUFs of different types of power Concentrated solar thermal 1 24 0,6 0 plants vary between 18% for hydropower and 19% for Solar PV Landfill 6 19 80 48 Industrial Waste 4 18 N/A N/A (lowest) to 87% for nuclear and 92% for Pig Manure (highest). In our Eco parks 6 17 56 37 example, only the power plants accounting for a minimum of 5% of the Forest Biomass 2 4 24 69 total installed capacity are considered. The CUFs are an important Waste treatments on farms 9 3 18 65 proxy to determine the function of different types of technologies: a low Total 580 12,440 46,456 CUF, paired with a non-renewable resource, usually means that the a CUF calculated as: (Electricity kWh/Installed Capacity kW) *(100/8760). plant has been used to cover peak demand (such as natural gas com- bined cycle plants, in this case). Similarly, high CUFs from non-re- newable resources imply that the power plant is producing electricity 5. Results and discussion used to cover baseload demand. For variable sources, the CUF alone does not provide a satisfactory indicator of the role played by the The model is built by aggregating lower-level structural elements electricity production system: the low CUF of solar panels, for example, into a higher level functional description. Mirroring Fig. 2, Fig. 3 shows does not mean they are used to cover peak demand, but is a reflection of the hierarchy of functions and structures considered in this case study – the intermittency of the PES. we will refer back to it consistently throughout the description of the various steps of the model, as we move from the description of tech- 4 nologies in relation to nexus patterns, to the description of the energy All sources of bottom-up data are included in the Appendix. 5 sector as a whole and its relation to other economic sectors. We refer to complex refineries as refineries which produce a large variety of petroleum products from crude oil, i.e. those that use thermal or catalytic cracking, or deep conversion processes. For more on refinery complexity, see 5.1. Production mix: structural and functional processors EIA (2012). 6 Defined as the ratio of the output of the plant in a year over the maximum The first step in the analysis is to build structural processors for electricity it could have produced in the same year. different energy technologies fulfilling the role of electricity, fuel and 435 L.J. Di Felice et al. Energy Policy 126 (2019) 431–443 Fig. 3. Hierarchy of structures and functions for the case study of Catalonia. heat generation. Structural processors can be defined either for specific of inputs and outputs is considered for each type: instances of a technology, e.g. a specific oil extraction platform, or for a type, e.g. for off-shore oil extraction platforms in Europe. The choice of 1. Flows from the technosphere: electricity and fuels; the types to be included in the taxonomy depends on the purpose of the 2. Flows from the biosphere: water; analysis. Since our purpose here is to give an overall description of the 3. Technosphere funds: land use, human activity (intended as labour) energy system, we present a generic structural description of the and power capacity; system, by including the main typologies of power plants, i.e. those that 4. Flows to the biosphere: GHG emissions. account for at least 5% of the total installed capacity. These are: Combined Heat and Power (CHP) plants, Nuclear Pressurised Water Therefore, each processor represents a pattern of the water-energy- Reactors (PWR), Natural Gas Combined Cycle (CC) plants, Small land-labour-climate nexus. We consider the water that is consumed Hydropower Plants (< 1 MW), Regular Hydropower Plants (> 1 MW), during the process, not including water withdrawn and then released. Pumped Hydroelectric Storage (PHS), On-Shore Wind Turbines and This means, for example, that the water flow input for hydropower Solar Photovoltaics (PV) for electricity production. For fuels, there is an plants represents the water evaporated during the process, and not the overlap between the type and the instance, as there is only one refining water passing through the dam. Labour, labelled as Human Activity, complex in Catalonia, so there is no need to build a typology (data is includes the working hours spent to produce the output (thus plant collected directly for the specific instance). manufacturing is not included) and does not include indirect jobs – In MuSIASEM, heat refers to either process heat or heating fuels. therefore, it describes the labour of operation & maintenance (O&M), Examples of heating fuels include refinery gases and petroleum coke. In and management overhead, accounted for in hours (h). Climate is ad- Catalonia, there are no processes producing singularly either of these, dressed by accounting for the GHG emissions that contribute to global therefore the functional element “heat generation”, which can be fur- warming according to the IPCC Fifth Assessment Report (AR5), and by ther broken down into “process heat generation” and “heating fuels converting them into CO2 equivalent (by using the characterisation generation”, is fulfilled by CHP plants and refineries. Once the struc- factors provided by the report). Land Use is intended as the area of land tural typologies are chosen, processors are built by defining a pattern of occupied by the power plant or installation. For hydropower, this inputs and outputs associated with each technology. The following set doesn’t include the occupation of water bodies. Since land use is not Fig. 4. Functional aggregation of electricity production processors. 436 L.J. Di Felice et al. Energy Policy 126 (2019) 431–443 discounted over the life of the processes, as the analysis has a yearly peak production allows us to comment on the role of renewables: since timescale, it is negligible for production of fuels and electricity (apart the electricity generated from renewable sources is outside human from renewables), as the majority of land used in the energy sector is control, it cannot be channelled into a desired function of baseload or during mining and extraction operations, and for the production of peak production. This is why a third processor of “intermittent gen- crops for biofuels. The choice of what to include in the processor and eration” is included in the analysis, as technologies relying on inter- how to define its inputs and outputs is not fixed. Moreover, definitions mittent sources have different metabolic characteristics than others, as can be tailored depending on the goal of the analysis and on who the well as different functionalities. Intermittent electricity can become information is being produced for (or with). functionally equivalent to either baseload or peak electricity when To compare and aggregate processors, all inputs are intensive, paired with storage technologies or back-up infrastructure, or both, by scaled by the main output – in this case, GWh of electricity for power increasing grid flexibility (Denholm and Hand, 2011; Steinke et al., plants and TJ of fuels for refineries. In this way, each process is de- 2013). scribed in relative terms (as a unitary operation). This kind of ab- Aggregating electricity generation into functional groups allows straction is useful in order to compare processes, for example to see comparing the differences in types of electricity generation, by building which type of power plant consumes more water per GWh produced. intensive functional However, in this framework, representations referring to unitary op- processors – a quantitative description of the functional group. erations are also scaled in terms of absolute values, in order to con- Table 4 shows how, for the same GWh of baseload, peak and inter- textualise them. This is where the functionality of different elements mittent power produced in Catalonia, a different pattern of flows and comes into play: a certain process may take up less land than others, but funds is needed.7 also produce a different type of energy carrier which is, in turn, used Some descriptive comments in relation to the nexus can be made: differently by a societal compartment to perform a specific task. This the production of a GWh of baseload electricity in Catalonia requires duality between the need for both intensive values referring to typol- double the amount of electricity than for peak and intermittent outputs; ogies and the need for extensive values contextualised to specific in- baseload electricity also requires a higher labour input than peak stances of energy systems is central to our approach. In this way, two electricity, however not higher than intermittent at 151 h per GWh non-equivalent descriptions of the system are considered simulta- produced. Water consumption is highest for peak electricity (more than neously. double what is needed for baseload), as well as GHG emissions, while The production of intensive descriptions of inputs and outputs be- the only type of electricity generation with a substantial land use re- comes problematic for co-generation processes, where the same profile quirement is intermittent, at 11 ha per GWh. These descriptive results of inputs refers to a profile of outputs. In our case this occurs in CHP may acquire value when contextualised in relation to policy questions. and refinement. To avoid allocation, a pattern of fixed output ratios is As mentioned previously, a combination of intensive and extensive associated with the typology. There are limits to this kind of simplifi- representations of nexus patterns, i.e. abstracted unitary representa- cation, but by avoiding allocation methods, the inherent entanglement tions and representations contextualised in relation to their size, is of patterns is explicitly addressed. useful to understand the behaviour of the energy system. Therefore, Table 3 shows an example of intensive data for the processor “On Table 5 includes inputs and outputs for functional electricity generation shore wind power”, where only four inputs are relevant. Similar tables groups, this time in an extensive format. This shows how, for example, are collected for each structural processor and are included in the intermittent electricity consume more human activity per GWh pro- Appendix. Then, a functional grouping of structural processors is per- duced than the other two types, but in extensive terms they are the formed, aggregating them based on the functionality of their outputs, sector which employs less human activity. The analysis for heat and and moving up the hierarchy (Fig. 4). For electricity, three functions are fuels is simpler for this case study, as the functional groups are not split identified: production of baseload electricity, production of peak elec- into further sub-functions. Table 6 shows the set of inputs and outputs tricity, and production of intermittent electricity. These three types of associated to the refinery process.8 electricity generation are well-known in engineering, where they are Having described the patterns of inputs and outputs required for the often split into further sub-categories such as middle-load (or load- production of electricity, heat and fuels, these can be aggregated into a following). However, in broader energy studies the distinction is rarely final pattern required for the whole energy sector, in what is re- made – see, for example, some of the 100% renewable energy studies presented as an energy sector processor (Fig. 5 and Table 7). This pro- published over the past decade, where all electricity is referred to as cessor reflects data which can be, in some cases, available from statis- being the same (Jacobson et al., 2015; Williams and Meisen, 2011; tical bodies, although often not to this grain of detail. However, Mason et al., 2010; Lund and Mathiesen, 2009). While physically (i.e. obtaining the set of data from bottom-up aggregation is key in under- structurally) all electricity is indeed the same, being a flow of electric standing the sets of different mechanisms and functionalities that have charge, it “usefuleness for society” depends on how, when and where it to be considered to explain the observed final pattern, and to build is generated and how, when and where it is used. scenarios. A sound understanding of the underlying metabolism leading Breaking down the electricity production curve into baseload and to the final behaviour of the energy sector is useful if one wants to modify such pattern, for example with the aim of reducing emissions or water use, while also understanding the effect that changes may have Table 3 on different nexus patterns at different levels. Table 7 represents the Structural processor (intensive) for "On shore wind power" Level −4. various inputs and outputs of the energy sector. The data in this case is extensive rather than intensive, therefore it is obtained by summing the Type: On-shore wind power extensive data of the three functional processors below it. The first Details Label Unit Source Internal flow input Electricity 22 MWh/GWh ICAEN (2012) 7 Internal flow input Fuels 0 GJ/GWh – This only includes “electricity generation”, and a full characterisation of the External flow input Water 0 m3/GWh – differences would have to include further steps in the energy chain (the full Internal fund input Human Activity 120 h/GWh Budia et al. (2013) sequential pathway). Internal fund input Power Capacity 470 kW/GWh ICAEN, (2012) 8 The inputs and outputs associated to the production of heat through CHP External fund input Land Use 12 ha/GWh Budia et al. (2013) and of heating fuels for refinery processes are accounted for in the structural External flow output GHG emissions 0 t CO2 eq./GWh – processors of CHP and of the refinery, therefore a separate accounting is not necessary. 437 L.J. Di Felice et al. Energy Policy 126 (2019) 431–443 Table 4 Comparison of intensive processors for baseload, peak and intermittent electricity Level −3. Details Label Unit Baseload Peak Intermittent Internal flow input Electricity MWh/GWh 42 23 20 Internal flow input Fuelsa GJ/GWh 216 0 0 External flow input Water m3/GWh 161 474 0 Internal fund input Human Activity (HA) h/GWh 111 57 151 Internal fund input Power Capacity (PC) kW/GWh 140 544 504 External fund input Land Use (LU) ha/GWh 0 0 11 External flow output GHG emissions t CO2 eq./GWh 69 321 0 a Fuels do not include the primary energy source, e.g. Uranium for PWR or natural gas for CHP. Table 5 Comparison of extensive inputs and outputs for functional electricity generation groups, and their relative share of the total Level −3. Label Unit Baseload (66%a) % Peak (27%b) % Intermittent (7%c) % Total Electricity GWh 1.26E + 03 79 2.71E + 02 17 6.00E + 01 4 1.59E + 03 Fuels TJ 6.41E + 03 100 0.00E + 00 0 0.00E + 00 0 6.41E + 03 Water 103 m3 4.79E + 03 46 5.69E + 03 54 0.00E + 00 0 1.05E + 04 Human Activity (HA) 103 h 3.29E + 03 74 6.86E + 02 15 4.51E + 02 10 4.43E + 03 Power Capacity (PC) MW 4.15E + 03 34 6.53E + 03 54 1.50E + 03 12 1.22E + 04 Land Use (LU) ha 1.81E + 02 1 2.37E + 02 1 3.31E + 04 99 3.35E + 04 GHG emissions t CO2 eq. 2.06E + 03 35 3.85E + 03 65 0.00E + 00 0 5.91E + 03 a,b,c The percentage given in brackets is in relation to the contribution of each functional group to Catalonia's total electricity production. Table 6 Extensive and intensive inputs and outputs of the refinery process. Source: PRTR Level −4. Details Label Unit intensive Value intensive Unit extensive Value extensive Internal flow input Electricity GWh/TJ 8.27E − 01 GWh 4.36E + 02 Internal flow input Fuels GJ/TJ 4.00E + 00 TJ 1.86E + 03 External flow input Water m3/TJ 1.40E + 01 m3 7.40E + 06 Internal fund input Human Activity h/TJ 4.00E + 00 h 1.95E + 06 External fund input Land Use ha/TJ 8.00E − 04 ha 4.43E + 02 External flow output GHG emissions t CO2 eq./TJ 4.00E + 00 t CO2 eq. 2.32E + 06 Internal flow output Total refinery output TJ 5.27E + 05 Fig. 5. Energy processor. 438 L.J. Di Felice et al. Energy Policy 126 (2019) 431–443 Table 7 Data for energy sector processor (including only generation of heat, fuels and electricity) Focal level. # on Fig. 5 Details Label Unit Value 1 Internal flow input Electricity GWh 2.03E + 03 2 Internal flow input Fuels TJ 8.27E + 03 3 External flow input Water 103 m3 1.79E + 04 4 Internal fund input Human Activity (HA) 103 h 6.38E + 03 5 Internal fund input Power Capacity (PC) MW 1.22E + 04 6 External fund input Land Use (LU) ha 3.39E + 04 7 External flow output GHG emissions t CO2 eq. 2.33E + 06 8 Internal flow output Electricity GWh 4.67E + 04 9 Internal flow output Fuels TJ 3.06E + 05 10 Internal flow output Heat TJ 1.37E + 05 Fig. 6. Link between energy processor and end uses. column of the table represents the labels shown in the energy sector of the coherence of the data across levels. processor of Fig. 5. Comparing the aggregated description of Catalonia's energy sector to top-down statistics, we see how the data presents some discrepancies 5.2. End uses: linking production to consumption (Table 9). We have not considered all steps of the energy sector, therefore it is expected for the aggregated bottom-up data to have lower The internal flow outputs, finally, can be allocated to different values than the statistical ones, as the table shows. Moreover, data sectors of society. Since part of what the energy sector produces is also specific to the energy sector is rarely available through statistical consumed within the sector itself, the representation linking the energy bodies, as it is usually aggregated with other sectors such as industry. sector processor to End Uses (Fig. 6) shows this loop. However, for values where data were available, bottom-up estimations For simplicity, only one loop back into the energy sector is re- fell within the right order of magnitude. In general, discrepancies be- presented in Fig. 6, while Table 8 shows how each energy carrier flow is tween aggregated and statistical data may be due to three types of in- consumed by different parts of society, including the energy sector. consistencies: (i) in the quality of statistical data; (ii) in the quality of bottom-up data; (iii) in the scaling relations moving from bottom-up to 5.3. Model validation and uncertainty top-down data. Further analyses of this kind could be used to discern trends in the inconsistencies, and determine the sources of error, flag- By aggregating bottom-up data into a higher level description, the ging possible systematic errors in statistical bodies. congruence between different sources of data can be checked. This is a A hierarchical organisation of the information characterising energy method often used by experts in their own operations, but rarely per- systems also allows us to open up a discussion on uncertainty. Looking formed in formal modelling since different types of data tend to be used back at Fig. 4, uncertainty may arise across different levels: there is for different types of analyses. In our framework, there is a forced re- uncertainty in the inputs and outputs depending on the uncertainty in lationship across levels, as the outputs of functional types, aggregated data sources and on the level of aggregation; moving up, uncertainties from technologies, must equal the inputs into end uses (taking into are associated to the combination of structural groups into functional account, of course, of trade and losses). This allows for a quality check ones; at the level of end uses, there is uncertainty in how many energy Table 8 End uses Level 1. Baseload electricity (GWh) Peak electricity (GWh) Intermittent electricity (GWh) Fuels (TJ) Heat (TJ) a Energy sector (ES) 1760 100 100 8010 57,800 Agriculture & Fishing (AF) 370 20 10 12,390 5320 Manufacture & Construction (MC) 14,880 1160 500 9640 85,080 Services & Government (SG) 7360 5880 1470 9300 13,940 Households (HH) 5330 4260 1070 10,430 41,620 Transport (TR)b 900 70 30 2,277,720 0 a data on fuel consumption of the energy sector does not include PES. b includes consumption of passenger vehicles. 439 L.J. Di Felice et al. Energy Policy 126 (2019) 431–443 Table 9 Comparison between aggregated data, and data obtained through statistical offices, for Catalonia in 2012. Focal level. Catalonia's energy sector, 2012 Unit Aggregated Statistical Source % difference GHG emissions t CO2 eq. 2.330E + 06 6.502E + 06 ICAEN 64 Water consumption 103 m3 1.790E + 04 N/Aa Labour 103 h 6.380E + 03 N/Aa Electricity consumption GWh 2.030E + 03 2.477E + 03 ICAEN 18 Fuels consumption TJ 8.270E + 03 1.829E + 04 ICAEN 55 Land use ha 3.390E + 04 N/Aa a Catalan government statistics do not disaggregate to this level of detail. carriers are consumed per sector, especially upon further disaggrega- OECD countries the energy sector can be seen as having the overall tion of electricity into baseload, peak and intermittent. The more we function of providing a supply of energy matching demand at all times. This develop a diversified characterisation of the expected relations over the expectation imposes local functions on its sub-components arranged to data, the more we can reduce the number of known unknowns (Knight, fulfil the sector's higher purpose, which in turn is aligned with functions 1921) relevant to the analysis. at the societal level (what the energy is used for in different economic compartments of society). By making this function explicit, it may be- 5.4. Policy relevance come contested. The strength and novelty of this energy model lie in its ability to 6. Conclusions facilitate informed discussions with stakeholders on policy relevant is- sues. The hierarchical mapping from primary energy sources, to the Ludwig and Walters (1985) have argued that simplified models are technologies harnessing them, to their function in society may help in more effective than large complex ones for policy decisions. On the guiding discussions of energy futures. The model carries a manageable other hand, the wisdom of Box (1979) reminds us that, ‘‘all models are amount of information across different domains, which can be com- wrong, but some are useful’’. Simple and effective models, rather than pressed or expanded depending on the policy question to be discussed. simplistic ones, are needed to deal with complex sustainability chal- Table 7 shows, for example, how the Manufacturing & Construction lenges. Iwasa et al. (1987) suggest that, in order to simplify complex sector in Catalonia is the one consuming the highest amount of baseload systems, attention should be turned to “techniques for choosing ap- electricity – this is because industries tend to operate 24 hours a day, propriate levels of aggregation”. This observation resonates with the 365 days a year. Over 80% of baseload electricity, on the other hand, is point made in the Introduction of this paper: when dealing with the produced by nuclear power (see Fig. 4). Therefore, when discussing a analysis of complex systems operating across different levels of orga- shift from nuclear power to renewable energy in Catalonia, it is im- nisation, their characteristics can only be observed by adopting dif- portant to consider the role that electricity from nuclear power plays in ferent scales. Thus, the quality of a representation based on a quanti- the industrial sector, and whether intermittent sources can be used to tative assessment depends on the ability of integrating non-reducible cover the same function, and if so at which social, capital (due to their observations. low CUF) and biophysical cost – for example, what are the implications The energy model introduced in this paper proposes a technique to for land uses needed for the renewable infrastructure? At what cost can choose appropriate levels of aggregation in the complex energy system, the grid become more flexible? On the other hand, the household sector based on the grouping of structural elements into functional ones. In has a high consumption of peak electricity, since demand in households this way, the hierarchical organisation of energy systems is used to varies throughout the day. In this case, flexible forms of demand-side simplify their representation and their relation to other nexus elements, management, integrated with intermittent generation, could be more while avoiding an excessive loss of relevant information. The strength successful. of the model lies across two domains. The first is in the domain of Moreover, Table 7 shows how electricity and thermal energy (heat energetics, contributing to a shift in the field towards inter- and fuels) are consumed differently by society. Structural changes can disciplinarity, by describing functional elements of the energy system partially be used to interchange the two, for example by using electric through nexus patterns. Secondly, at the science-policy interface, the vehicles to reduce fuel consumption in the transport sector. However, model may be used as a heuristic tool to inform decision-making, not all functions covered by thermal energy can be replaced by elec- moving away from a paradigm of science speaking truth to power, and tricity – a further disaggregation of end uses into specific processes towards the co-production of knowledge among scientists, policy- could show, for example, how many of the fuels consumed in the makers and other stakeholders. transport sector are for road vehicles and how many are for navigation and aviation, where electricity may not be a suitable substitute. In the Manufacturing & Construction sector, similarly, some processes such as Acknowledgments smelting furnaces used to make iron require a thermal input which renewable sources struggle to provide. Trade-offs across levels may also The authors gratefully acknowledge research funding from the emerge: a technological change, such as switching to a power plant European Union's Horizon 2020 research and innovation programme consuming less water per output, may have positive effects on local under grant agreement No. 689669 (MAGIC). The Institute of ecosystems but produce unwanted effects on nexus patterns at higher Environmental Science and Technology (ICTA) has received financial hierarchical levels, or be constrained by which changes are perceived to support from the Spanish Ministry of Economy and Competitiveness, be desirable at the level of end uses. through the “María de Maeztu” program for Units of Excellence (MDM- These examples show how the recognition of the importance of 2015-0552). This work reflects the authors’ view only; the funding functionality and its inclusion in the hierarchical organisation of the agencies are not responsible for any use that may be made of the in- model is useful in informing policy questions across nexus domains. In formation it contains. 440 L.J. Di Felice et al. Energy Policy 126 (2019) 431–443 Appendix A Table A.1 Structural Processors Level −4. Details Label Value Unit Source Type: Nuclear PWR Internal flow input Electricity 45 MWh/GWh ICAEN (2012, n.d.) Internal flow input Fuels 90 GJ/GWh Diaz-Maurin and Giampietro (2013) External flow input Water 50 m3/GWh ANAV (2012) Internal fund input Human Activity 65 h/GWh ANAV (2012) Internal fund input Power Capacity 131 kW/GWh ICAEN (2012, n.d) External fund input Land Use 0.004 ha/GWh NEI (2015) External flow output GHG emissions 0 – – Type: Natural Gas Combined Cycle (CC) Internal flow input Electricity 26 MWh/GWh ICAEN (2012, n.d.) Internal flow input Fuels 0 GJ/GWh PRTR, n.d. External flow input Water 650 m3/GWh Dones et al. (2007) Internal fund input Human Activity 61 h/GWh Internal fund input Power Capacity 493 kW/GWh ICAEN (2012, n.d.) External fund input Land Use 0.022 ha/GWh NEI (2015) External flow output GHG emissions 460 t CO2 eq./GWh Dones et al. (2007) Type: Combined Heat & Power (CHP) Internal flow input Electricity 33 MWh/GWh ICAEN (2012, n.d.) Internal flow input Fuels 748 GJ/GWh ICAEN (2012, n.d.) External flow input Water 630 m3/GWh Dones et al. (2007) Internal fund input Human Activity 305 h/GWh ACOGEN (2010a) Internal fund input Power Capacity 178 kW/GWh ICAEN (2012, n.d.) External fund input Land Use 0.015 ha/GWh ACOGEN (2010b) External flow output GHG emissions 363 t CO2 eq./GWh Dones et al. (2007) Type: Small hydro (< 1 MW) Internal flow input Electricity 14 MWh/GWh ICAEN (2012, n.d.) Internal flow input Fuels 0 GJ/GWh Flury and Frischknecht (2012) External flow input Water 0 m3/GWh Flury and Frischknecht (2012) Internal fund input Human Activity 25 h/GWh Benchmark from private communication with an engineer working at Endesa Internal fund input Power Capacity 355 kW/GWh ICAEN (2012, n.d.) External fund input Land Use 0.006 ha/GWh Dones et al. (2007) External flow output GHG emissions 0 t CO2 eq./GWh Dones et al. (2007) Type: Regular hydro (> 5 MW) Internal flow input Electricity 145 MWh/GWh ICAEN (2012, n.d.) Internal flow input Fuels 0 GJ/GWh Flury and Frischknecht (2012) External flow input Water 29 m3/GWh Dones et al. (2007) Internal fund input Human Activity 50 h/GWh Benchmark from private communication with an engineer working at Endesa Internal fund input Power Capacity 760 kW/GWh ICAEN (2012, n.d.) External fund input Land Use 0.023 ha/GWh Dones et al. (2007) External flow output GHG emissions 0.4 t CO2 eq./GWh Dones et al. (2007) Type: Pumped hydro storage (PHS) Internal flow input Electricity 145 MWh/GWh ICAEN (2012, n.d.) Internal flow input Fuels 0 GJ/GWh Flury and Frischknecht (2012) External flow input Water 160 m3/GWh Dones et al. (2007) Internal fund input Human Activity 50 h/GWh Benchmark from private communication with an engineer working at Endesa Internal fund input Power Capacity 550 kW/GWh ICAEN (2012, n.d.) External fund input Land Use 0.026 ha/GWh Dones et al. (2007) External flow output GHG emissions 0.4 t CO2 eq./GWh Dones et al. (2007) Type: On-shore wind power Internal flow input Electricity 22 MWh/GWh ICAEN (2012, n.d.) Internal flow input Fuels 0 GJ/GWh – External flow input Water 0 m3/GWh – Internal fund input Human Activity 120 h/GWh Budia et al. (2013) Internal fund input Power Capacity 468 kW/GWh ICAEN (2012, n.d.) External fund input Land Use 12 ha/GWh Budia et al. (2013) External flow output GHG emissions 0 t CO2 eq./GWh – Type: Solar PV Internal flow input Electricity 0 MWh/GWh ICAEN (2012, n.d.) Internal flow input Fuels 0 GJ/GWh – External flow input Water 0 m3/GWh – Internal fund input Human Activity 440 h/GWh APPA (2012) Internal fund input Power Capacity 856 kW/GWh ICAEN (2012, n.d.) External fund input Land Use 3 ha/GWh APPA (2012) External flow output GHG emissions 0 t CO2 eq./GWh – Tarragona Refinery Internal flow input Electricity 0.8 MWh/TJ PRTR, n.d. Internal flow input Fuels 4 GJ/TJ PRTR, n.d. External flow input Water 14 m3/TJ PRTR, n.d. (continued on next page) 441 L.J. Di Felice et al. Energy Policy 126 (2019) 431–443 Table A.1 (continued) Details Label Value Unit Source Internal fund input Human Activity 4 h/TJ PRTR, n.d. Internal fund input Power Capacity N/A kW/TJ PRTR, n.d. External fund input Land Use 0.000841 ha/TJ PRTR, n.d. External flow output GHG emissions 4410 t CO2 eq./TJ PRTR, n.d. Note: “Water” is intended here as water consumed or evaporated throughout the process (not water throughput that is then released); Inputs and outputs scaled per GWh of CHP also have a corresponding fixed pattern of heat being generated; Inputs and outputs scaled per TJ in the refinery refer to TJ of total products (fuels + heat + other products). References Future: Multi-Scale Integrated Analysis of Societal and Ecosystem Metabolism. Routledge, pp. 360. Giampietro, M., Aspinall, R.J., Ramos-Martin, J., Bukkens, S.G.F., 2014. Resource ACOGEN, 2010a. Valoración de los beneficios asociados al desarrollo de la cogeneración Accounting for Sustainability Assessment: The Nexus between Energy, Food, Water en España. and Land use. Routledge, pp. 250. ACOGEN, 2010b. Guia de la cogeneracion. pp. 57–66. González-López, R., Giampietro, M., 2017. Multi-scale integrated analysis of charcoal Ahl, V., Allen, T.F.H., 1996. Hierarchy Theory: A Vision, Vocabulary, and Epistemology. production in complex social-ecological systems. Front. Environ. Sci. 5, 54. https:// Columbia University Press, New York. doi.org/10.3389/fenvs.2017.00054. ANAV, 2012. Anav memoria anual. González-López, R., Giampietro, M., 2018. Relational analysis of the oil and gas sector of APPA, 2012. Estudio del Impacto Macroeconómico de las Energías Renovables en España. Mexico: Implications for Mexico's energy reform. Energy 154, 403–414. Appa 144. Grenon, M., 1978. Welmm, a Systems Approach To Energy Resources 2. Aragão, A., Giampietro, M., 2016. An integrated multi-scale approach to assess the per- Howarth, C., Monasterolo, I., 2016. Understanding barriers to decision making in the UK formance of energy systems illustrated with data from the Brazilian oil and natural energy-food-water nexus: the added value of interdisciplinary approaches. Environ. gas sector. Energy 115, 1412–1423. https://doi.org/10.1016/j.energy.2016.06.058. Sci. Policy 61, 53–60. https://doi.org/10.1016/J.ENVSCI.2016.03.014. Bazilian, M., Rogner, H., Howells, M., Hermann, S., Arent, D., Gielen, D., Steduto, P., Howells, M., Hermann, S., Welsch, M., Bazilian, M., Segerström, R., Alfstad, T., Gielen, D., Mueller, A., Komor, P., Tol, R.S.J., Yumkella, K.K., 2011. Considering the energy, Rogner, H., Fischer, G., van Velthuizen, H., Wiberg, D., Young, C., Roehrl, R.A., water and food nexus: towards an integrated modelling approach. Energy Policy 39, Mueller, A., Steduto, P., Ramma, I., 2013. Integrated analysis of climate change, land- 7896–7906. https://doi.org/10.1016/j.enpol.2011.09.039. use, energy and water strategies. Nat. Clim. Chang. 3, 621–626. https://doi.org/10. Box, G.E.P., 1979. Robustness is the strategy of scientific model building. In: Launer, R.L., 1038/nclimate1789. Wilkinson, G.N. (Eds.), Robustness in Statistics. Academic Press, New York, pp. ICAEN, 2012. Production of electrical energy in Catalonia [WWW Document], n.d. URL 201–236. 〈http://icaen.gencat.cat/es/energia/estadistiques/resultats/anuals/produccio_ Budia, Á., Carbajal, S., Ceña, A., Franco, S., Mora, Á., Morante, M., Mesa, P., Polo, L., energia/〉 (accessed 30 November 2016). Rosique, K., Simonot, E., Willstedt, H., 2013. Eolica 2013, referencia de sector. Asoc. ICAEN Statistics [WWW Document], n.d. URL 〈http://icaen.gencat.cat/es/energia/ Empresa. Eolica. https://doi.org/10.1017/CBO9781107415324.004. estadistiques/〉 (accessed 15 November 2016). Cairns, R., Krzywoszynska, A., 2016. Anatomy of a buzzword: the emergence of ‘the Iwasa, Y., Andreasen, V., Levin, S., 1987. Aggregation in model ecosystems. I. Perfect water-energy-food nexus' in UK natural resource debates. Environ. Sci. Policy 64, aggregation. Ecol. Modell. 37, 287–302. https://doi.org/10.1016/0304-3800(87) 164–170. https://doi.org/10.1016/j.envsci.2016.07.007. 90030-5. Cleveland, C.J., 1987. Biophysical economics: historical perspective and current research Jacobson, M.Z., Delucchi, M.A., Bauer, Z.A.F., Savannah, C., Chapman, W.E., Cameron, trends. Ecol. Modell. 38, 47. M.A., Cedric, A., Chobadi, L., Erwin, J.R., Fobi, S.N., Owen, K., Harrison, S.H., Cottrell, F., 1955. Energy and society: the relation between energy, social changes, and Kwasnik, T.M., Lo, J., Yi, C.J., Morris, S.B., Moy, K.R., Neill, P.L.O., Redfern, S., economic development, 13209th Ed. McGraw-Hill. Schucker, R., Sontag, M.A., Wang, J., Weiner, E., Yachanin, A.S., 2015. 100% Clean Denholm, P., Hand, M., 2011. Grid flexibility and storage required to achieve very high and Renewable Wind, Water, and Sunlight (WWS) All- Sector Energy Roadmaps for penetration of variable renewable electricity. Energy Policy 39, 1817–1830. https:// 139 Countries of the World By 2050. Draft 1–61. doi.org/10.1016/j.enpol.2011.01.019. Keairns, D.L., Darton, R.C., Irabien, A., 2016. The energy-water-food nexus. Annu. Rev. Diaz-Maurin, F., Giampietro, M., 2013. A “ Grammar” for assessing the performance of Chem. Biomol. Eng. 7, 239–262. https://doi.org/10.1146/annurev-chembioeng- power-supply systems: comparing nuclear energy to fossil energy. Energy 49, 080615-033539. 162–177. https://doi.org/10.1016/j.energy.2012.11.014. Knight, F.H., 1921. Risk, uncertainty and profit. Hart, Schaffner and Marx, New York. Dodds, P.E., Keppo, I., Strachan, N., 2015. Characterising the evolution of energy system Koestler, A., 1968. The Ghost in the Machine. MacMillan, New York. models using model archaeology. Environ. Model. Assess. 20, 83–102. https://doi. Koestler, A., 1978. Janus: a Summing Up. Hutchinson, London. org/10.1007/s10666-014-9417-3. Kovacic, Z., 2017. Investigating science for governance through the lenses of complexity. Dones, R., Bauer, C., Bolliger, R., Burger, B., Heck, T., Röder, A., Institut, P.S., Futures 91, 80–83. Emmenegger, M.F., Frischknecht, R., Jungbluth, N., Tuchschmid, M., 2007. Life Cycle Kurian, M., 2017. The water-energy-food nexus: trade-offs, thresholds and transdisci- Inventories of Energy Systems: Results for Current Systems in Switzerland and other plinary approaches to sustainable development. Environ. Sci. Policy 68, 97–106. UCTE Countries. Ecoinvent report No. 5. Paul Scherrer Institut Villigen, Swiss Centre https://doi.org/10.1016/J.ENVSCI.2016.11.006. for Life Cycle Inventories, Dübendorf, CH. Lam, J.C.K., Walker, R.M., Hills, P., 2014. Interdisciplinarity in sustainability studies: a EIA, 2012. Petroleum refineries vary in level of complexity [WWW Document]. URL review. Sustain. Dev. 22, 158–176. https://doi.org/10.1002/sd.533. 〈https://www.eia.gov/todayinenergy/detail.php?Id=8330〉 (accessed 2 October Lotka, A.J., 1922. Contribution to the energetics of evolution. Proc. Natl. Acad. Sci. 8. 2018). Louie, A.H., 2010. Robert Rosen's anticipatory systems. Foresight 12, 18–29. https://doi. Endo, A., Burnett, K., Orencio, P., Kumazawa, T., Wada, C., Ishii, A., Tsurita, I., Taniguchi, org/10.1108/14636681011049848. M., 2015. Methods of the water-energy-food nexus. Water 7, 5806–5830. https://doi. Ludwig, D., Walters, C.J., 1985. Are age-structured models appropriate for catch-effort org/10.3390/w7105806. data? Can. J. Fish. Aquat. Sci. 42, 1066–1072. https://doi.org/10.1139/f85-132. European Commission, 2018. Modelling tools for EU analysis [WWW Document]. URL Lund, H., Mathiesen, B.V., 2009. Energy system analysis of 100% renewable energy 〈https://ec.europa.eu/clima/policies/strategies/analysis/models_en〉 (accessed 2 systems – the case of Denmark in years 2030 and 2050. Energy 34, 524–531. https:// July 2018). doi.org/10.1016/j.energy.2008.04.003. European Commission, 2016a. Clean Energy For All Europeans. Brussels. Mason, I.G., Page, S.C., Williamson, A.G., 2010. A 100% renewable electricity generation European Commission, 2016b. EU reference scenario 2016: energy, transport and GHG system for New Zealand utilising hydro, wind, geothermal and biomass resources. emissions trends to 2050. European Commissionhttps://doi.org/10.2833/9127. Energy Policy 38, 3973–3984. https://doi.org/10.1016/j.enpol.2010.03.022. Flury, K., Frischknecht, R., 2012. ESU-services Ltd. Life Cycle Inventories of Hydroelectric NEI, 2015. Land Requirements for Carbon-Free Technologies, pp. 4–5. Power Generation. ESU - Serv. Odum, H.T., 1971. Environment, power and society. Environ. Power Soc. Funtowicz,S., Ravetz, J., 1990. Uncertainty and quality in science for policy. Dordrecht. Ostwald, W., 1907. The modern theory of energetics. Monist 17, 481–515. https://doi. Georgescu-Roegen, N., 1971. The Entropy Law and the Economic Process. Harvard org/10.2307/27900060. University Press, Boston. Parra, R., Di Felice, L.J., Giampietro, M., Ramos-Martin, J., 2018. The metabolism of oil Giampietro, M., 2018. Anticipation in agriculture. In: Poli, R. (Ed.), Handbook of extraction: a bottom-up approach applied to the case of Ecuador. Energy Policyd 122, Anticipation. Springer, Cham. https://doi.org/10.1007/978-3-319-31737-3_23-1. 63–74. https://doi.org/10.1016/J.ENPOL.2018.07.017. (Online ISBN: 978-3-319-31737-3). Pereira, Â.G., Funtowicz, S. (Eds.), 2015. Science, philosophy and sustainability: the end Giampietro, M., Allen, T.F.H., Mayumi, K., 2006. The epistemological predicament as- of the Cartesian dream. Routledge. sociated with purposive quantitative analysis. Ecol. Complex. 3 (4), 307–327. PRTR - Espana, inventario de instalaciones [WWW Document], n.d. URL 〈http://www. Giampietro, M., Mayumi, K., Sorman, A.H., 2012. The Metabolic Pattern of Societies: prtr-es.es/Informes/InventarioInstalacionesIPPC.aspx〉 (accessed 12 January 2017). Where Economists Fall Short. Routledge. Ravetz, J., 1971. Scientific Knowledge and its Social Problems. Oxford University Press. Giampietro, M., Mayumi, K., Sorman, A.H., 2013. Energy Analysis for a Sustainable Rayner, S., 2012. Uncomfortable knowledge: the social construction of ignorance in 442 L.J. Di Felice et al. Energy Policy 126 (2019) 431–443 science and environmental policy discourses. Econ. Soc. 41 (1), 107–125. scholarship and proposing a social science research agenda. Energy Res. Soc. Sci. 1, Ringler, C., Bhaduri, A., Lawford, R., 2013. The nexus across water, energy, land and food 1–29. https://doi.org/10.1016/j.erss.2014.02.003. (WELF): potential for improved resource use efficiency? Curr. Opin. Environ. Sustain. Steinke, F., Wolfrum, P., Hoffmann, C., 2013. Grid vs. storage in a 100% renewable 5, 617–624. https://doi.org/10.1016/j.cosust.2013.11.002. Europe. Renew. Energy 50, 826–832. https://doi.org/10.1016/j.renene.2012.07. Rosen, R., 1991. Life Itself. Columbia University Press, New York. 044. Rosen, R., 2000. Essays on Life Itself. Columbia University Press, New York. Stirling, A., 2015. Developing ‘Nexus Capabilities’: towards transdisciplinary methodol- Saltelli, A., Giampietro, M., 2017. What is wrong with evidence based policy, and how ogies. can it be improved? Futures 91, 62–71. White, L.A., 1943. Energy and the evolution of culture. Am. Anthropol. 45, 335–356. Simon, H., 1962. The architecture of complexity. Proc. Am. Philos. Soc. 106, 467–482. https://doi.org/10.1525/aa.1943.45.3.02a00010. https://doi.org/10.1080/14759550302804. Whyte, L.L., Wilson, A.G., Wilson, D., 1969. Hierarchical Structures. Elsevier, New York. Smil, V., 2010. Energy Transitions: History, Requirements, Prospects. Praeger, Santa Williams, N., Meisen, P., 2011. Can Spain reach 100% renewable energy by 2020? Renew. Barbara. Energy. Sovacool, B.K., 2014. What are we doing here? Analyzing fifteen years of energy 443