Applied Energy 298 (2021) 117210 Contents lists available at ScienceDirect Applied Energy journal homepage: www.elsevier.com/locate/apenergy The declining performance of the oil sector: Implications for global climate change mitigation Michele Manfroni a, *, Sandra G.F. Bukkens a, Mario Giampietro a, b a Institut de Ciència i Tecnologia Ambientals, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain b Institució Catalana de Recerca i Estudis Avançats (ICREA), Pg. Lluís Companys 23, 08010 Barcelona, Spain H I G H L I G H T S • The declining performance of the oil sector is studied with relational analysis. • Aging fields and unconventional sources may hinder climate change mitigation. • Emissions per barrel are expected to increase by 6–26% in 40 years. • The emission overhead is comparable to emissions of entire EU economic sectors. • The option space of potential changes is explored by running “what if” simulations. A R T I C L E I N F O A B S T R A C T Keywords: This article presents a relational analysis of the performance of the petroleum sector in the context of climate Oil depletion change mitigation. The oil sector is described as a complex network of transformations carried out by structural Unconventional oils and functional elements, exploiting different types of crude oils. Energy carrier requirements and emissions of Fossil transition viable sequential pathways of extraction and refining are assessed and scaled across different levels of organi­ CO2 emissions Relational analysis zation, using the concept of metabolic processor. Based on the analysis of seventy-one oil fields around the world Climate change mitigation - about 25% of global production - we provide a diagnostic analysis of the current state and explore possible scenarios simulating the progressive aging of conventional oil sources and an increasing exploitation of un­ conventional crudes. Results show how future oil exploitation will be more energy intensive, entailing an in­ crease of emissions per barrel in the range of 6–26% over the baseline, depending on the simulation. Under the existing policy frameworks and international pledges, this increase will translate into an amount of extra CO2 comparable to entire European economic sectors. Implications of our findings for future energy policies are discussed and the need to complement Integrated Assessment Models (IAMs) with more robust methodologies is emphasized. It is concluded that the declining performance of the oil sector could potentially undermine the plausibility of global low-carbon aspirations. 1. Introduction transition period, the energy sector will have to provide not only a sufficient supply of net energy for the daily functioning of the economy, Climate change is on the front burner of the political agenda, given but also for the required extra investments in low-carbon technologies its huge potential impacts on the biosphere and the global economy and infrastructures [7,8]. This (largely fossil) energy “overhead” trans­ [1,2]. A rapid decarbonization process is therefore advocated almost lates into an overhead of CO2 emissions, which will determine the bio­ unanimously by all national governments and international bodies physical pay-back time of the energy investment - i.e., the time required [3,4]. However, in the short term, the energy input required for such a by alternative energy sources to compensate for the extra CO2 emissions radical transformation will have to be supplied by the current energy accumulated for building and operating green infrastructures and matrix, which still consists predominantly of fossil fuels (about 90% of technologies [9,10]. the world energy consumption in 2019) [5,6]. Hence, during the Given the ubiquitous carbon lock-in in the primary and secondary * Corresponding author at: Edificio ICTA-ICP (Z), UAB Campus de Bellaterra, 08193 Bellaterra, Spain. E-mail address: michele.manfroni@uab.cat (M. Manfroni). https://doi.org/10.1016/j.apenergy.2021.117210 Received 13 October 2020; Received in revised form 26 May 2021; Accepted 27 May 2021 Available online 23 June 2021 0306-2619/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). M. Manfroni et al. Applied Energy 298 (2021) 117210 Fig. 1. The oil sector as a metabolic network. Extraction typologies are represented in grey, refining typologies in blue. Components are related by the functional entailment of producing fuels (red) at the superior scale. For the sake of simplicity, only functional elements are shown. production sectors, it is unlikely that we will be able to radically change oil production comes from giant fields, whose major discoveries (in our energy matrix and substantially reduce CO2 emissions in the near terms of production capacity) date back to the period 1940–1980. The future [11,12]. Looking at current fossil fuel end-uses, we find that: (i) progressive depletion of those fields makes the oil extraction more transportation is almost 100% petroleum based; (ii) industry is heavily difficult [24] and Enhanced Oil Recovery (EOR) techniques of different dependent on petrochemical feedstocks and fuels for essential processes complexity are increasingly needed to keep up the production. EOR like plastic, cement, steel and glass production; (iii) agriculture entails more energy carriers (heat, electricity and fuels) and water use completely depends on ammonia and machinery; (iv) electricity pro­ (to keep the reservoir in pressure and pump out fluids with increasing duction is dominated by coal, natural gas and uranium [13,14]. To this water-to-oil ratios), to produce the same amount of net supply. The picture we have to add the sunk-cost of the already existing technical consequence is a decline of the Energy Return On Investment (EROI) and capital [15], buildings [16] and infrastructures [17], which generates an higher emissions per unit of net supply [25,26]. The process of aging important fossil fuel path dependency that is not easy to change. In also affects the crude quality, with older fields often producing heavier addition, rapid population growth in some developing countries and the and dirtier oils [27]. Therefore, considering an average age of currently strive for better living conditions worldwide are likely to further in­ depleted fields of around 80 years [28], the effects of this aging phe­ crease fossil energy use and CO2 emissions. Hence, not surprisingly, both nomenon will become fully evident between 2020 and 2060, which is fossil energy consumption and emissions continue to rise despite the the critical time window for climate change mitigation, according to the Paris agreement, UNFCC and IPCC international meetings [18,19]. IPCC [29]. At the same time, the worldwide ongoing “uncon­ At the global level, oil still represents the first primary energy source ventionalization” process of oil supplies affects the infrastructures and (accounting for 33% of the total primary sources) [5]. Rather than a technologies needed for extraction, transportation and refining. More transition away from fossil fuels, two interrelated drivers of change complex drilling and deeper refining of unconventional sources increase dominate the oil sector: (i) the progressive depletion (aging) of con­ the energy intensity and the related emissions of the changing oil in­ ventional oil fields [20]; and (ii) a “fossil transition” from conventional dustry [30]. to unconventional sources, entailing a complexification of the supply The effects of these two drivers challenge the plausibility of a rapid system [21]. The two processes are intrinsically entangled: while con­ renewable energy transition and of climate change mitigation scenarios. ventional oil production is progressively declining, an array of new oils, Therefore, before making ambitious plans for a radical decarbonization such as oil sands, tight oil, new heavy and extra-heavy oils, ultra-deep of the global economy in 2050 [29,31], it is important to assess the waters oils and oil shales is projected to fill the gap. From 2005 on­ wider impacts of the declining performance of the oil sector on our low- wards, the world production has been sustained by US tight oil, which carbon future aspirations. accounts for 60% of the global increase in supply. Conventional crude is Currently commonly used methodological approaches, such as Life- expected to account for only 60% of liquid-fuel supply by 2040, Cycle Assessment (LCA), strive to produce accurate carbon intensity compared to 80% in 2012 [22]. assessments for specific oil products (or net energy units) to inform Based on the evidence reported in [20,23], about 60% of the global policy-making [32,33]. For instance, LCA assessments have been 2 M. Manfroni et al. Applied Energy 298 (2021) 117210 Fig. 2. Examples of scaling: sequential pathway and parallel composition of structural and functional elements. The sequential pathway represents a material entailment (directed edge): the output of extraction is the input for refining. The parallel composition represents a hierarchical entailment (undirected edges) and interrelates two or more levels: the elements are tied together by the same function at higher scales. NB. Elements can be either structural or functional, depending on the scale of observation: they can be functional processes in relation to higher levels, or structural types (‘blueprints’) for elements at lower levels. provided for unconventional extraction [34], refining [35], and the The purpose of this paper is to gain better insight into the effects of entire well-to-wheel oil supply chain [28]. However, the findings of the aging and “unconventionalization” processes of the global oil sector, these LCA analyses are valid only within the (narrow) context defined by so as to support the assessment of the option space for plausible decar­ the assumptions of the analyst. Indeed, LCA results are highly sensitive bonization pathways. By using relational analysis, we first characterize to system boundaries definition, efficiency benchmarks and the taxon­ the biophysical performance of the global oil sector across multiple omy of functional units [36]: choosing different boundaries or allocation scales and dimensions, using data from seventy-one oil fields and asso­ criteria among different products leads to significantly different assess­ ciated refineries world-wide, representing about 25% of global pro­ ments for the same fuel pathway [37,38]. In practice, these assessments duction. Subsequently, we simulate a progressive aging of existing oil generate questionable ‘optimal’ solutions for narrow policy domains (e. fields and an increasing exploitation of unconventional oil sources, and g., the California Low Carbon Fuel Standard, LCFS), rather than a multi- assess the related changes in the consumption of energy carriers and CO2 dimensional knowledge space for an informed discussion. emissions. For each simulation, we focus on the change in emissions and The same narrow optimization scope, but using highly-aggregated, energy requirement per barrel, where the barrel is a weighted average of general equilibrium models, is pursued by the IPCC’s assessments of viable sequential pathways of oil extraction and refining. We then use climate change mitigation pathways [29,39]. These models represent the forecasted level of oil supply worldwide of the IEA “Stated Policies the current state-of-the-art of the Integrated Assessment Modelling Scenarios” [49] to provide a rough estimate of the absolute magnitude of (IAM) for climate change mitigation options, and push for innovation the increase in CO2 emission that may be expected from the fossil and efficiency towards a clean energy transition [40-42]. Nonetheless, transition in the oil sector. these IAMs of energy and CO2 emissions pathways have been widely The rest of this paper is structured as follows. Section 2 explains the criticized [43,44], in particular the basic assumption of energy- methodology and data sources. Section 3 describes the results of the abundance [45]. Several studies, using different economic [46] and analyses. Section 4 discusses the strength and shortcomings of the geological models [47,48] to relate CO2 emissions and supply pro­ approach. Section 5 provides some policy indications and concludes. jections, have shown that a decline in fossil fuel production will constrain mitigation options and future climate impacts. Surprisingly, 2. Materials and methods the structural implications of the fossil transition within the oil sector (aging of conventional oil fields and the transition to unconventional 2.1. Relational analysis of the oil sector oils) on required energy investments and GHG emissions, and the resulting consequences for climate change mitigation pathways, are yet Our analytical framework is grounded in Rosen’s theory of relational to be comprehensively investigated. analysis (originally born as ‘relational biology’) and represents a 3 M. Manfroni et al. Applied Energy 298 (2021) 117210 relatively novel approach for characterizing the performance of complex “transportation” has been included in the step of “extraction”, as a flat systems, such as the energy sector [27,50]. overhead. In brief, in relational analysis an energy system is described as a metabolic network of structural and functional elements (Fig. 1). The study starts from an identification of functional elements for which it is 2.2. Taxonomy of structural and functional elements possible to identify expected relations. Only in a second phase, the characteristics of structural elements are studied [51,52]. The distinc­ 2.2.1. Oil fields and extraction tion between structural and functional strictly depends on the scale of The taxonomy of oil fields (typologies of structural elements) and the observation and the distinction between tangible (instances) or notional corresponding extraction processes (typologies of functional elements) (types) elements considered in the representation. Structural elements has been defined based on the following criteria: are ‘tangible elements’ such as oil fields or refineries - e.g., the Ghawar field, Fig. 2. These structural elements are associated with structural 1. API Gravity − 3 categories: (i) light, (ii) medium, (iii) heavy oils; typologies sharing a common set of metabolic attributes (inputs and 2. Water content − 2 categories: (i) low water, (ii) watery; outputs profile) - e.g., “Light watery offshore” fields in Fig. 2 or “Medium 3. Location – 2 categories: (i) onshore, (ii) offshore. conversion” refineries in Fig. 1. Different structural elements can be combined together in order to express a functional element – i.e., a The combination of these criteria results in twelve different typol­ ‘notional element’ described in terms of expected profiles of inputs and ogies of oil fields and corresponding extraction processes (illustrated in outputs. In this case, the metabolic characteristics of functional elements Fig. 1). To these, an additional four types were added, namely: (i) - that are not tangible – can be assessed by calculating the characteristics “Fracking” and (ii) “Ultra-Deep” crudes, both of which are characterized of the particular mix of structural types making up the functional type. by distinct technologies of extraction with specific biophysical re­ The use of notional elements is essential to describe the performance of quirements and environmental burdens; (iii) “Light and medium the energy system. Examples of notional (functional) elements in Fig. 1 depleted” and (iv) “Heavy depleted” oils, both characterized by a long are “Light watery offshore” extraction or “Deep conversion – coking” history of exploitation and evident signs of aging (depletion). refining. Both structural and functional elements can be described in In this work, only fracking and ultra-deep crudes are referred to as quantitative terms as profiles of inputs and outputs through the use of unconventional oils. Given the ambiguity in the use of the label “un­ metabolic processors (see section 2.3). The quantification of structural conventional” [22,53], this semantic choice was made to distinguish types is derived from the observations of equivalence classes of given them from the categories “heavy” and “heavy depleted”. For the purpose instances of structural elements, while functional elements are observed of this analysis, heavy and depleted fields, that, under specific circum­ as nodes in the network and calculated as combinations of the charac­ stances (like advanced EOR techniques or very high viscosity) can be teristics of mixes of structural elements (or functional elements at higher considered unconventional oil, are accounted for separately, in order to levels). enrich the taxonomy and appreciate the effect of time on a greater va­ The relations over the nodes in the network (the identities of “meta­ riety of crudes. bolic processors” in the jargon of relational analysis) can be scaled in two different ways (see Fig. 2): 2.2.2. Refineries and refining Four different typologies of refining processes (functional elements) 1. Parallel composition, when different profiles describing different have been considered to classify the refineries (facilities, structural el­ classes of functional elements (i.e., different types of extraction or ements) in our data base: refining) are composed, according to their relative contribution (i.e., percentage in the mix), into an aggregated metabolic profile of the 1. Hydroskimming conversion; functional element defined at a higher level of analysis. This is 2. Medium conversion; illustrated by the undirected edges between nodes across levels in 3. Deep conversion - coking; Fig. 2. 4. Deep conversion – hydrocracking. 2. Sequential pathway, when different structural or functional elements are linked by a material entailment, i.e., the output of a node is the The choice of coking versus hydrocracking depends on the refiner. input of the next node. This is illustrated by the directed edge be­ Hydrocracking maximizes middle distillates, while coking maximizes tween ‘Extraction’ and ‘Refining’ (on the same level) in Fig. 2. light distillates. The choice entails different input requirements and products outcome. As detailed in Section 2.4, all the deep conversion First, structural elements at level n-3 are combined together in a refining in our simulations is assumed to be coking, even if both typol­ parallel composition to form structural types at level n-2. In other words, ogies have been considered and assessed within the space of viable different instances (fields) of oil extraction, such as the “Ghawar field”, sequential pathways. Note that the complexity of refining increases with are combined together into different types of extraction, such as “Ultra- API gravity and sulfur content of the crude oil: more energy and tech­ Deep”. Then, those structural types are combined into the functional nical intensive processes are required to transform the heavier and element “Extraction” at level n-1. At that level, different functional ele­ dirtier oils. ments – i.e., Extraction and Refining – can be combined into a sequential pathway determining the characteristics of a functional element – i.e., 2.2.3. Identifying viable sequential pathways “Fuels Supply” – defined at the superior level n. At the end of this chain, a Using the taxonomy of typologies of the structural and functional mix of different final oil products (see appendix, Table A4) is made elements defined in the previous subsections, different sequential available for end-uses in the other sectors of society and in the energy pathways were constructed according to the metabolic network shown sector itself (internal loop of energy carriers to produce energy carriers, in Fig. 1. Note that not all possible combinations of extraction and i.e., ‘energy-for-energy’). The composition of this final mix of oil prod­ refining processes are viable: even if technologically possible, heavy oils ucts depends on the relative composition and specific identities of the usually are not processed in hydroskimming refineries due to the low structural and functional elements in the metabolic network. yields of high quality (and high economic value) products such as gas­ In the present study, a two-stage sequential pathway has been oline and diesel. In the same way, the economic viability of low and considered: (i) extraction: extraction processes, in-site upgrading and/or medium complexity refineries is increased with the selective processing transformations and transport to refineries; and (ii) refining. To simplify of light and medium oils. Hence, we have used the following criteria for the analysis and for data availability reasons, the sequential stage of generating sequential pathways: 4 M. Manfroni et al. Applied Energy 298 (2021) 117210 Fig. 3. Metabolic processor for the functional element “Fracking”. • Light, medium and ultra-deep oils are associated with hydro­ economically viable products determined by societal demand). The use skimming or medium conversion refining depending on their sulfur of metabolic processors for the characterization of the metabolic pattern content; of social-ecological systems has been described in detail in [55]. • Tight oil from fracking always goes to hydroskimming refineries; Quantification starts out from data that refer to the characteristics of • Heavy oils can be refined by medium or deep conversion plants, specific instances of structural elements, i.e., the observed fields and either by coking or hydrocracking facilities; refineries. The inputs and outputs for the metabolic processors used in • For the purpose of refining, light, medium and heavy depleted oils this study include: are assumed to be equivalent (in physical–chemical properties) to their respective non-depleted variant and linked to same refining • For extraction processors: typology. 1. Net consumption of natural gas (NG) and refinery fuel gas (RFG) (MJ/bbl); The resulting viable sequential pathways are characterized in the 2. Net and indirect (associated with imports) consumption of electricity results section (see Section 3.1 and Table 3). (MJ/bbl); Once viable sequential pathways have been identified, it is possible 3. Net and indirect consumption of diesel (MJ/bbl); to characterize their relative importance (contribution) in the supply 4. On-site and indirect CO2 emissions (kgCO2eq/bbl); chain. In this way, the analyst can scale up the assessment of the re­ 5. Daily Production Capacity (bbl/day). quirements of inputs (mix of energy carriers) and the CO2 emissions produced for the supply of one barrel of products to the level n. This • For refining processors: quantitative characterization has been done for the actual situation 1. Net consumption of natural gas and refinery fuel gas (MJ/bbl); (diagnostic) and for scenarios. Results are reported in Section 3. 2. Net consumption of electricity (MJ/bbl); 3. Net consumption of diesel (MJ/bbl); 2.3. The metabolic processor 4. Net consumption of coke (MJ/bbl); 5. On-site CO2 emissions (kgCO2eq/bbl); The metabolic processor is an analytical tool organizing quantitative 6. Daily refining capacity (bbl/day); information in the form of a data array. It can be interpreted as an 7. Product slate (%). extended production function in that it defines an expected (functional) or observed (structural) profile of inputs and outputs associated with a For a detailed explanation of the data processing based on the OCI specific process, such as oil extraction or refining [54]. Metabolic pro­ Climate Index data base within our relational procedure, see the cessors allow for a quantitative representation of the relations between Appendix. structural and functional elements across levels (Fig. 2). In this way, we Fig. 3 illustrates the processor of the functional element “Fracking” can quantify the functional units of the oil sector by bridging the re­ (extraction). This simplified illustration shows the main features of the lations between processors, either looking at sequential pathways organization of the information in a data array. The technosphere, determined by a material entailment or by a parallel composition, when shown in the upper right part of Fig. 3, refers to the conversions (end considering a functional entailment (different elements expressing the uses) of flows taking place under human control. This includes: (i) the same function at the higher level). The choice of a mix of functional and input requirement of energy carriers (natural gas, diesel, electricity and structural elements depends both on biophysical (set of typologies of oil coke) to produce oil (the energy-for-energy loop); and (ii) the output of fields and refineries available) and socio-economic constraints (set of barrels of crude (that then becomes the input consumed by refineries). 5 M. Manfroni et al. Applied Energy 298 (2021) 117210 Table 1 Relative contribution of extraction typologies to the total oil supply in the current situation (2018) and in the four simulations. Unconventional oils are reported in green. Note that the internal energy loop for natural gas is usually closed at the to characterize the current supply system at the global level, data on the level of the extraction processor, while diesel, electricity and coke are relative contribution of different crudes from [56] have been used, coming from ‘outside’, i.e., either from a higher level of the oil sector referring to the year 2017. (after refining) or from imports (nonetheless, for some oil fields also In 2017, light, medium and heavy oils represented 31.7%, 54.8% and natural gas is imported). The information in the technosphere is relevant 13.5% respectively of the total crudes extracted [56]. Tight oil (entirely for the economic and technical viability of the process and to assess the light) from fracking accounts for 30% of the North American production correct output of oil products (Table A4). In the lower left part of Fig. 3, (2018), that represents 12% of the world total extraction. We assumed inputs and outputs that are relevant for studying processes outside of that the amount from ultra-deep fields (>3000 m) worldwide equals that human control (in the biosphere) are shown. These include: (i) the of tight oils, equally divided between light and medium crudes (this is supply capacity of oil (and associated gas) stocks (i.e., oil fields); and (ii) relevant in order to associate the ultra-deep crudes with the appropriate the sink capacity of primary biophysical flows (CO2 emissions in this refining). Therefore, overall unconventional oils (fracking and ultra- study). deep) are assumed to represent 7% of global production [57]. The same structure of inputs and outputs is used to quantify relations In order to assign a proper relative share to the other 14 categories, between structural and functional elements across all levels of the entire we used production capacity from OPGEE v1.1_draft_e assessment [28]. metabolic network, thus allowing the generation of an integrated The formula used for the calculation is a weighted average on the pro­ quantitative representation. Note that metabolic processors can be duction capacity of each typology: expressed either as extensive processors (based on a given size of the ∑s ∑r ∑n i=1 PCi,j,k flows) or unitary processors (benchmarks, per unit of throughput) [54]. A= k ∑j=1 s Further explanations on the construction of metabolic processors for the k=1 PC k characterization of the performance of the oil sector are available in Where PC is the production capacity, n = number of oil fields of the same [27,50]. typology, r refers to sub-typologies of extraction considering water content and location (4), s are the sub-typologies of extraction consid­ 2.4. Data sources and assumptions ering API gravity (3). The contribution from depleted fields was rounded to 10% (“Light and medium depleted” fields 9% and “Heavy depleted” 2.4.1. Diagnosis of the current situation fields 1% on the total). The final result is reported in the column ‘current The relational analysis has been implemented using raw data from situation’ in Table 1, showing the relative contributions of all extraction seventy-one oil fields and associated refineries world-wide, representing typologies and taken as baseline reference for generating simulations. 25% of global production, from [28]. Specifically, OPGEE_v1.1_draft_e is the data source for extraction processors, while data for refining pro­ 2.4.2. Anticipation of possible scenarios cessors are from PRELIM v.1.2 (https://www.ucalgary.ca/energy-tech Starting from the current situation defined above (the actual nology-assessment/open-source-models/prelim). In addition, in order composition of the global supply), four different simulations were 6 M. Manfroni et al. Applied Energy 298 (2021) 117210 Table 2 data from [56]. Ultra-light and light & sweet oils (73.5% of light types) Refining shares associated with extraction typologies. are associated with hydroskimming technology, medium and heavy Associated refining Hydroskimming Medium Deep sweet crudes are converted in medium conversion facilities while the types conversion conversion sour heavy crudes in deep refining plants. Tight oil from fracking goes Light low water 74% 26% 0% 100% to hydroskimming; light, ultra-deep oils are supposed to be 50% onshore light and 50% medium and follow the rules above, light and medium Light low water 74% 26% 0% depleted are 100% medium converted and, finally, heavy depleted are offshore 100% led to deep conversion. The refining shares have been kept con­ Medium low water 0% 100% 0% onshore stant throughout all scenarios. Table 2 shows a schematic representation Medium low water 0% 100% 0% of the refining assumptions. offshore Note that all deep refining is assumed to be done with coking facil­ Light watery onshore 74% 26% 0% ities. For our purposes, distinction between coking and hydrocracking Light watery offshore 74% 26% 0% has not been considered, since we are not taking into account changing Medium watery 0% 100% 0% onshore demand patterns. However, a ‘hydrocracking’ scenario would be more Medium watery 0% 100% 0% emission intensive. Coking optimizes gasoline production. Its energy offshore carrier input requirements as well as emissions are lower than with Heavy low water 0% 19% 81% hydrocracking technology, which is better for diesel production. onshore Heavy low water 0% 19% 81% offshore 2.4.3. “Stated Policies Scenario” Heavy watery 0% 19% 81% By combining the unitary quantification per barrel, obtained from onshore the scaling of metabolic processors into viable sequential pathways, Heavy watery 0% 19% 81% with the relative extraction and refining shares, it becomes possible to offshore Ultra-Deep 37% 63% 0% appreciate the actual consumption of energy carriers in the system and Fracking 100% 0% 0% CO2 emissions, as well as products output per barrel in each scenario. Light and medium 0% 100% 0% The size of the system (barrels produced per day/year) can be scaled up depleted to the magnitude required. In this work, we refer to the International Heavy depleted 0% 0% 100% Energy Agency’s (IEA) World Energy Outlook 2019 “Stated Policies Sce­ nario” [49]: the global oil production is projected to be 105 million explored (see Table 1). In all simulations and also in the ‘current situ­ barrels per day in 2030 and 106 million in 2040. We use the former ation’, the same overall oil supply is guaranteed and kept constant at quantity to scale up the analysis to the global perspective, both in the 105 million barrels per day – from IEA’s forecast “State Policies Sce­ current scenario and simulations. This was done to specifically assess the nario” (2030) - but the relative composition of the various structural and qualitative effects of aging and “unconventionalization” against a fixed functional elements (extraction, refining) changes because of aging and (quantitative) supply. the fossil transition toward unconventional sources. The assumptions for the simulations are: 2.4.4. Data representation: Normalized chromatic intensity Maintaining data disaggregated is important for preserving a multi- • Simulation #1: 50% of the oil production from light and medium, dimensional information space (e.g., processors’ benchmarks for low-water, onshore fields shifts to the respective watery types, and different typologies of energy carriers) and identifying useful patterns. 50% of the light and medium, watery, onshore fields becomes However, data proliferation represents a significant challenge for the depleted (“Light and medium depleted” fields). visualization of quantitative information. For this reason, in several ta­ • Simulation #2: the same as Simulation#1, but now 100% of the oil bles (Tables 3–5), we have used Normalized Chromatic Intensity (NCI), production from the selected types changes functional category; that is gradients of color intensities, to facilitate the recognition of • Simulation #3: same as Simulation #2, but in addition all heavy patterns [58]. The creation of NCI representations follows three steps: production types move to the heavy depleted type. first, identifying the maximum and minimum values for the selected • Simulation #4: ultra-deep and fracking cover 40% of total produc­ variable over the set of data; second, calculating the range of values for tion (20%+20%), at the expense of light and medium, low water and the variable (difference between maximum and minimum value of the onshore pathways. series); and third, assigning proportional intensities of color for the in­ termediate values in relation to its normalized distance to the extremes The first three simulations focus on the progressive decrease in the of the interval (maximum color intensity for highest values and mini­ quality of existing oil fields (low-water fields becoming watery and mum intensity for lowest ones). In this way, chromatic visualizations of depleted), which, based on the current evidence, is expected to take patterns over the observed dimensions are obtained and outliers in the place during the next 40 years, as already stated in the introduction. The data set are easily identified. fourth simulation looks at the potential effects of a sensible increase in unconventional oil production, probably larger than what is considered 3. Results plausible [57], but still useful to highlight possible trends. The simula­ tions are implemented by changing the relative contributions of the 3.1. Typologies of production pathways and diagnostic of current different functional types of extraction to the global production. As situation stated earlier the overall supply remains constant. The characteristics (unitary representation) of each type of extraction processor (its iden­ Table 3 summarizes the consumption of energy carriers and CO2 tity) is based on the current situation, and are assumed to remain con­ emissions per barrel of oil products supplied across current, viable stant. This assumption is discussed in Section 4.2. sequential pathways of the oil sector. Table 4 reports the arrays of oil Information about the relative contributions of the specific sequen­ products expected as outcome from each refining (functional) typology. tial (extraction-refining) pathways inside the supply system is needed to Further details about mixes of oil products per sequential pathway are scale up the information across the metabolic network both in the cur­ provided in the Appendix, Table A3, while in Tables A1 and A2 rent situation and in the scenarios. These assessments are based on the extraction and refining processes are separately reported. From a unitary perspective, heavy and depleted pathways are, not 7 M. Manfroni et al. Applied Energy 298 (2021) 117210 Table 3 Viable sequential pathways of the oil sector (extraction - refining), requirement of energy carriers (energy-for-energy loop) and CO2 equivalent emissions per barrel, unitary description. 8 M. Manfroni et al. Applied Energy 298 (2021) 117210 Table 4 Relative composition of the oil products per barrel for different (functional) typologies of refining processors. Abbreviations: LHE: Liquid Heavy Ends. Table 5 Requirement of energy carriers (energy-for-energy loop) and CO2 equivalent emissions per barrel, weighted by the relative contribution of sequential pathways (Table 1) in the current situation. Unitary description. Fig. 4. Global annual absolute CO2 emissions if oil supply would be equally distributed between sequential pathways - i.e., if every supply share in the columns of Table 1 were 6.25% - under the IEA’s (World Energy Outlook 2019) “Stated Policies Scenario” 2030 production assumption (105 million barrels per day). 9 M. Manfroni et al. Applied Energy 298 (2021) 117210 Fig. 5. Global annual absolute CO2 emissions for the current relative composition of sequential pathways of oil supply under the IEA’s (World Energy Outlook 2019) “Stated Policies Scenario” 2030 production assumption (105 million barrels per day). Fig. 6. Requirement of energy carriers per barrel of oil products for the four simulations provided. Fig. 7. Comparative average increase per barrel of oil products in the requirement of energy carriers and CO2 emissions between the current situation (baseline) and the four simulations provided. surprisingly, the most energy and emission intensive (Table 3 and not a major reason for concern because of their limited capacity of Fig. 4). But when enlarging the scale of assessment, considering the production. When studying them using the “extensive processor” global production capacities of each typology, the picture changes. In (aggregated quantities of flows), they do not play a substantial role, fact, heavy oils, although most polluting in unitary terms (Table 3), are neither in delivering fuels nor in consuming energy carriers and emitting 10 M. Manfroni et al. Applied Energy 298 (2021) 117210 Fig. 8. CO2 equivalent emissions per end-use in EU27 compared with the marginal emission increase for the four simulations performed. CO2 (Table 5 and Fig. 5). simulation considered (Fig. 7). The majority of those emissions are due Comparing “Heavy low water onshore” with “Light low water to increased consumption of natural gas and electricity. Again, the dif­ onshore” pathways on a per barrel basis, the former shows a + ference between extensive and intensive characteristics is important 311–562% emission range per barrel (depending on the typology of here. Even if the use of natural gas and electricity as energy carriers is refining). However, the bulk of the global production (Table 1) is pro­ less emission intensive per unit of energy delivered (compared with vided by light and medium, low-water and onshore fields, which diesel or coke, for example), it is the size of the supply of oil that matters. together cover up to 45% of the worldwide crude supply. They account Incidentally, there are serious doubts about the possibility of main­ for the largest consumption of energy carriers and related emissions in taining oil production at the current level, since productivity per well the energy sector (Table 5 and Fig. 5). Currently, all the heavy and drops substantially with reservoir depletion and the related increase in depleted pathways together account for only 28 kgCO2eq per barrel water content (waterflooding is a common secondary oil recovery (‘GHG emission’ column, Table 5), equal to the sum of the light and technique to keep production high in spite of the deterioration of field medium low-water onshore pathways, at 29 kgCO2eq/bbl. Nonetheless, quality) [61,62]. heavy and depleted pathways are relevant considering their possible Compared to the effects of aging (simulations 1-3), a shift to un­ growth in the overall mix in the near future (“too big to ignore” [59]). conventional pathways (simulation 4) appears less dramatic in terms of These results show the importance of maintaining non-equivalent emission increases (Fig. 7). Although the use of diesel almost doubles in representations of the oil sector to better inform decision makers by this simulation, in absolute terms this increase is much smaller than the triangulating intensive (unitary benchmarks per types) and extensive increase in natural or refinery gas consumption observed in simulations descriptions (the extensive processors indicating the size of instances of 1-3, hence the relative emission increase per barrel is smaller. However, types). Considering only emissions per barrel as if every sequential other relevant aspects, not explored in the present work, should be pathway had the same importance, misses the implications of the rela­ considered, such as technical difficulties in the extraction of ultra-deep tive size and may lead to policy decisions with limited efficacy, as is the fields [63] and their quick pace of decay, requiring a continuous capital case with “smart taxes” on high-carbon oils [60]. investment in power capacity for drilling new wells [64]. The simulations indicate that the main drivers of increasing emis­ sions are: (i) the increase in natural gas consumption because of aging; 3.2. Results of simulations and (ii) the increase in diesel consumption for the “unconventionaliza­ tion” of oil sources. Natural gas is used heavily in extraction to generate In this section, it is anticipated what will happen to the energy carrier the heat needed for EOR techniques: waterflooding and most of tertiary requirements and CO2 emissions of the energy sector when the relative EOR technologies employ natural gas to pump fluids (water, steam or contribution to the oil supply from depleted and watery oil fields pro­ gases) and generate steam or direct heat to reduce the viscosity in the gressively increases (simulations 1–3) or when the energy sector reservoir in order to extract crude. Deep refining is especially intensive increasingly relies on unconventional sources (simulation 4). in natural gas use due to the various hydrocracking, hydrotreating, Energy carrier requirements per barrel of oil for the four simulations desulfurization and reforming facilities needed to treat heavy oils and is reported in Fig. 6. Energy carriers considered are natural gas (NG) and make the fuels compliant with environmental laws. Diesel is used mainly refinery fuel gas (RFG), electricity, diesel and coke. The relative increase in ultra-deep extraction, due to the (almost exclusively) offshore loca­ in energy carrier requirements and CO2 emissions of the 4 simulations tion and geological nature of the reservoir (depth). In Tables A1 and A2 compared to the current situation (baseline) is shown in Fig. 7. of the appendix, more detailed quantitative results are provided. In simulations 1–3, watery and depleted pathways progressively gain A last observation is due on the quality of the oil products supplied. weight in the oil supply mix at the expense of light and medium low- In our scenarios the relative composition of the mix of oil products water onshore ones. Since the former pathways are more intensive in supplied to society remains fairly constant (see Appendix, Table A4). terms of consumption of energy carriers and emissions, each barrel This indicates that the quality of the final output is not significantly produced will cause higher emissions compared to the current situation. affected by aging or increased reliance on unconventional sources. To maintain the current output of hydrocarbon fuels, a 6–26% increase Given the large size of the global oil sector, these simulations are in CO2 emissions per barrel should be expected depending on the 11 M. Manfroni et al. Applied Energy 298 (2021) 117210 relevant for a responsible discussion about climate change. Depending 1. Diagnostic mode, that is the description of the current metabolic on the simulation, the absolute increase in GHG emissions is between pattern of the oil production sector across levels and dimensions of 251 and 747 MtCO2eq/year for a total global production of 105 million analysis; barrels per day. These extra emissions are in the same order of magni­ 2. Anticipation mode, that is the creation and evaluation of future tude as those of the agricultural or manufacturing & construction sector scenarios by changing the wiring inside and across the sequential of the EU27 and, in simulation 3, nearly comparable to the emissions of pathways or by assuming different technical coefficients for the the EU transportation system [65] (see Fig. 8). structural elements in the nodes. Thus, if we want to meet the energy demand of 105 million barrels per day by 2030 under the current energy policy frameworks and in­ Defining the constraints on both the biophysical (production pro­ ternational pledges worldwide (i.e., the “Stated Policies Scenario”), we cess) and socio-economic side (fuel demand), the coherence and sus­ need to compensate for the declining performance of the oil sector by tainability of fuel consumption patterns can be identified and, playing reducing emissions elsewhere. The “unconventional revolution” [66] with the relationships found, anticipated. will aggravate, not solve the problem: even if the unconventional barrel is slightly less emission intensive than crude from depleted pathways, 4.2. Synchronic versus diachronic analysis to study aging emissions per barrel increase compared to the current situation, espe­ cially due to intensive diesel consumption in ultra-deep extraction. In this study, the changes in time of the functional types of the oil Renovated abundance of unconventional oil resources will exacerbate sector (diachronic data of the CO2 emissions per barrel) are inferred by the absolute GHG emissions and the pressure on the climate system. combining: (i) the benchmarks of structural types observed at a given point in time (synchronic data of the technical characteristics of struc­ 4. Discussion tural elements); and (ii) anticipated changes in the relative mix of these structural types. This procedure overlooks the effects of possible tech­ This section highlights the strength and the novelty of the proposed nological improvements taking place at the level of specific processes (e. relational analysis and addresses the limitations of the current study g., pumping volumes of flows). Nonetheless, when considering aging with regard to (i) synchronic versus diachronic analysis to study aging; and “unconventionalization” as drivers of change, the relational analysis (ii) input/output variables considered in the metabolic process; (iii) the is robust given the comparative relevance of: (i) a possible increase in scope of the study. the technical efficiency within a given structural typology; and (ii) an important change in the mix of structural typologies within the given 4.1. The strength and novelty of relational analysis functional types. Indeed, technical improvements in the efficiency of the technological devices operating at the oil wells (structural elements) are Relational analysis provides a robust and versatile tool for biophys­ much less relevant than changes in the mix of structural elements (e.g., ical assessment in that it can define typologies of extraction and refining the relative share of “Heavy watery onshore” or “Light low-water processes through the generation of unitary metabolic processors based onshore” oil fields) determining the functional typology. The extrac­ on benchmarks. tion of oil from depleted fields comes with huge amounts of (contami­ Typologies and their unitary processors can be easily compared to nated) produced water (including formation and injection water and each other and can be used to construct scenarios in which the relative added chemicals). Low-water fields typically show water-to-oil bench­ contributions of typologies vary. More precisely, different types of oil marks of less than 1/4 (80% oil vs 20% water). With progressive production and refining can be analyzed in functional terms in relation depletion, these proportions may easily reach up to 4/1 (20% oil vs 80% to the role they play inside the energy sector even when considering water). In such a situation, any potential efficiency improvements per energy sectors of different size (either observed or simulated). Indeed, barrel in the individual structural types (at best in the order of 20–30%) the representation of relational networks based on metabolic processors are more than offset by the huge increase in the extracted volume to be offers the analyst flexibility. Using this tool, it is possible to integrate the processed in the new (anticipated) mix determining the functional type. unitary (notional) description with the extensive description (based on The same holds true for unconventional extraction due to the advanced observation i.e., statistical data), building a bridge across dimensions techniques needed to extract difficult-to-access (ultra-deep and tight oil) and levels of analysis and allowing for the consideration of the effect of crudes. Regarding refining, while ultra-deep and fracking crudes are not size and the spatial location of the energy transformations. Biophysical problematic for refiners, more complex facilities are needed to process costs, environmental pressures and the supply of products delivered to dirtier or stickier oil from aged fields. society can be scaled across levels and quantified using simultaneously For these reasons, we believe that a reliable estimate of the changes different metrics. Two main advantages of these features are: in time of the overall performance of the oil sector can be obtained by simulating only the changes in the profile of structural elements, 1. It allows the analyst to generate a multi-criteria performance space assuming current technical benchmarks to be constant. A study by Parra characterizing the supply of oil products using ad-hoc indicators (e. et al. [27], analyzing the aging of Ecuadorian oil reserves, validates g., specific input/output ratios) related to costs, efficiency and these assumptions. Undoubtedly, for future analyses, it would be pref­ pollution; erable to use time series of benchmarks from the same oil fields so as to 2. The multi-scale, multi-dimensional description across different levels include the combined effect of changes in technical characteristics of of analysis can be tailored to the specific problem the analyst wants structural types and functional characteristics determined by the mix of to study. For instance, the representation of performance can be fit to structural types. However, historical series at the required level of data relevant levels of analysis, e.g., a single stage of the process, a specific detail are generally not easily available. functional element, or the whole supply system. Finally, as shown in the current study, relational analysis can be 4.3. Input/output variables included in the metabolic processors meaningfully used in two ways: The current study focused on energy carrier consumption and direct CO2 emission in the oil sector to study the ‘aging trap’ of the energy transition. The economic – i.e., capital and operational expenditures – 12 M. Manfroni et al. Applied Energy 298 (2021) 117210 and technical investments – i.e., kW of power machinery and quality of the aging and complexification of the oil sector, in the form of increas­ technology employed – were not included in the metabolic representa­ ingly watery and depleted fields and a progressive switch to uncon­ tion. The economic dimension, albeit relevant, is out of the scope of the ventional sources, lead to increased energy carrier consumption and CO2 biophysical representation presented in this study. Regarding the power emissions per barrel produced by the oil industry. This phenomenon is capacity dimension, the issue is the lack of available data. To the best of the result of efforts to merely maintain the current size of fuels output. our knowledge, only oil companies have data at that level of detail, and This “oil aging trap” is often overlooked – e.g., in the IPCC decarbon­ those are not publicly shared. However, in future and more refined ization scenarios based on IAMs – but has important implications for the analyses, it would be important to include economic and technical in­ international efforts to contain climate change in the next decades to puts too. In this way, one could study the effect of the continuous in­ come. crease of investments in new technical capital, needed to overcome the There are several potential applications of the findings of this paper. challenges of aging (e.g., EOR) and the exploration and extraction of First, the findings indicate that focusing exclusively on the emissions per unconventional fields. In fact, rapidly decaying fracked wells [67], and barrel for specific oil typologies – e.g., energy policies based on LCAs – is sophisticated ultra-deep extraction technologies [63,68] entail not only potentially misleading and may lead to ineffective carbon taxes and more direct emissions, but also important energy requirements and emission trade policies, because these assessments overlook the relative emissions embodied in the installed power capacity. A methodological importance of the various sequential pathways in the overall oil mix. discussion of how to integrate power capacity into the processor’s rep­ Second, addressing the main drivers of increased CO2 emissions in resentation has been provided by Diaz-Maurin [69]. the oil sector, namely natural gas for more complex drilling and deeper refining, should be among the priority concerns of energy policies. Reducing the use of natural gas (in the form of hydrogen) in complex 4.4. The scope of the study refining – needed to supply high quality fuels from low quality primary oils - could be useful for climate change mitigation. Hence, energy The simulations presented in this paper are not meant to predict the policies that stimulate burning cleaner fuels and the upgrading rate of future, but as a possible way to explore option spaces for energy tran­ refineries should be carefully evaluated in the face of (i) possible trade- sition by gaining insights into the consequences of the aging and com­ off between pollution and CO2 emissions; and (ii) the creation of addi­ plexification of the global oil sector. The main aim of the work is to tional fossil fuels path dependency in the economy. provide a useful analytical framework for reflexive governance, capable Third, rather than focusing all attention on the use of renewables to of complementing the mainstream approach, which exclusively focuses replace fossil fuel, more research is needed on the possibility of using on “optimization and control”. renewables in fossil fuel extraction. Technologies such as Solar The selected taxonomy of extraction and refining processors is Enhanced Oil Recovery could prove useful for reducing emissions and, at implicitly based on the historical and common accounting scheme the same time, by-pass the main problems of large-scale, societal adopted by oil companies, not on a cluster analysis of the biophysical implementation of renewable sources, namely intermittency and profiles (input and output of primary and secondary flows) of the oil storage. fields (see Appendix, Figs. A1, A2, A3). This choice has been made The operational framework presented in this paper has been devel­ because, in the world of oil, every field is a complex object, difficult to oped specifically for governance purposes. It allows analysts and policy geologically characterize and exploit by engineers: uncertainty is the makers to identify relations and trade-offs between relevant and con­ norm and every producer employs specific sets of technologies [70]. flicting concerns associated with the production of energy carriers in Consequently, the biophysical profiles characterizing the extraction societies still heavily dependent on fossil fuels, but willing to reduce CO2 processors (but not the refining ones, that are less dispersed between emissions. Our analysis illustrates the potentiality of the approach in instances of the same type) show high variability and every clustering providing a useful integrated characterization of the diversity of pro­ constructed upon requirements of natural gas, diesel, electricity, coke cesses of energy conversions taking place in the oil sector, and in iden­ and CO2 emissions could easily be contested. Hence, in agreement with tifying priority areas for further research. Decisions of policy-makers [28], we selected API gravity, water content, location and depth as should not be based on simplistic indicators aimed at indicating optimal relevant attributes because of their implications for the capital in­ solution (often used out of context). Missing the big picture because of a vestments in extraction and refining technologies. reductionist approach and a simplistic thinking that we can solve However, the choice of the taxonomy of structural and functional climate change simply through technological innovations and business elements and the choice of inputs and outputs in the processors can al­ models will force us to run the Red Queen’s Race: “It will take all the ways be contested within the relational framework. This predicament is running we can do, just to keep in the same place” …but only for a unavoidable in any analysis of complex systems across different levels limited time. and dimensions. Relational analysis makes this choice transparent. In fact, we are not claiming that our numbers are correct or indisputable, CRediT authorship contribution statement nor that we are providing an accurate assessment of the increase in the emissions/supply ratio of oil products due to the aging and complex­ Michele Manfroni: Conceptualization, Methodology, Formal anal­ ification of the oil sector. What we showed is that the “oil aging trap” is a ysis, Visualization, Writing - original draft, Writing - review & editing. relevant aspect of the problem of climate change mitigation that should Sandra G.F. Bukkens: Writing - review & editing. Mario Giampietro: be (better) addressed in the narratives about decarbonization of the Conceptualization, Methodology, Writing - review & editing, economy (and in the making of IAMs). Grounding the representation of Supervision. the oil sector in a metabolic perspective allows the analyst to obtain a broader view of the reality observed. Declaration of Competing Interest 5. Conclusions The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence The modern energy matrix still relies predominantly on fossil fuels the work reported in this paper. and the transition to renewable energy sources is expected to take de­ cades at least. In this paper we have shown that, while the clock ticks, 13 M. Manfroni et al. Applied Energy 298 (2021) 117210 Acknowledgements Units of Excellence (CEX2019-000940-M). The paper has greatly benefited from the valuable comments and suggestions of the anony­ Michele Manfroni gratefully acknowledges financial support from mous reviewers. Michele Manfroni also wants to thank Fabrizio Ciatti the Catalan administration/AGAUR (Grant number: 2018 FI_B 00313). for his valuable suggestions. This work reflects the authors’ view only; The Institute of Environmental Science and Technology (ICTA) has the funding agencies cannot be held responsible for any use that may be received financial support from the Spanish Ministry of Science, Inno­ made of the information it contains. vation and Universities, through the “María de Maeztu” program for Appendix • Additional data on the functional processors and sequential pathways referred to in the main text Fig. A1. Visualization of the ratio “Standard Deviation/Average” of the natural gas consumption (in blue) and GHG emissions (in orange), calculated over the specific oil fields (instances of extraction, structural elements) belonging to the same extraction (functional) typology. Standard deviation is consistently high for both flows for most extraction (functional) typologies, entailing high data variability for the characterization of oil fields. *benchmarks refer to a single field. Fig. A2. Visualization of the ratio “Standard Deviation/Average” of the natural gas consumption, GHG emissions and oil products (gasoline, jet fuel, diesel, coke, Liquid Heavy Ends), calculated over the refineries (instances of refining, structural elements) belonging to the same refining (functional) typology. *PRELIM assumption: coke is not used in refining; **PRELIM assumption: fuel oil is completely converted to high-value products (gasoline or diesel) in refinery types other than hydroskimming. 14 M. Manfroni et al. Applied Energy 298 (2021) 117210 Fig. A3. CO2 emissions per barrel supplied, per oil field (e.g., “Nigeria Obagi”, structural element in black) and per extraction typology (e.g., “Fracking”, functional element in red). Note that large standard deviations are cause by the outliers in the upper part of the figure, however most of the fields are consistent with the benchmark of their reference type. Table A1 Metabolic profiles (i.e., identities) of extraction processors: consumption of energy carriers and CO2 emissions. Extraction processors MJ/bbl kgCO2eq/bbl NG and RFG Electricity Diesel Coke GHG emissions Light low water onshore 90 4 7 0 25 Light low water offshore 176 7 13 0 32 Medium low water onshore 199 3 8 0 46 Medium low water offshore 33 2 5 0 20 Light watery onshore 425 38 6 0 78 Light watery offshore 523 8 5 0 74 Medium watery onshore 357 13 14 0 76 Medium watery offshore 174 11 4 0 31 Heavy low water onshore 1570 1 1 0 143 Heavy low water offshore 331 5 13 0 55 Heavy watery onshore 218 21 20 0 52 Heavy watery offshore 71 8 0 0 20 Ultra-Deep 199 6 40 0 51 Fracking 108 2 17 0 59 Light and medium depleted 431 35 7 0 55 Heavy depleted 1937 10 1 0 127 Table A2 Metabolic profiles of refining processors: consumption of energy carriers and CO2 emissions. Refining processors MJ/bbl kgCO2eq/bbl NG and RFG Electricity Diesel Coke GHG emissions Hydroskimming 297 10 0 0 17 Medium conversion 454 16 0 25 31 Deep conversion - coking 974 32 0 49 71 Deep conversion - hydrocracking 1278 38 0 41 94 15 M. Manfroni et al. Applied Energy 298 (2021) 117210 Table A3 Viable sequential pathways of the oil sector: supply of oil products per barrel, unitary description. Sequential pathways Types of extraction Associated types of refining Gasoline Jet Fuel Diesel Fuel Oil Coke LHE Light low water onshore Hydroskimming 29% 23% 10% 10% 0% 28% Medium conversion 36% 19% 18% 0% 0% 26% Light low water offshore Hydroskimming 29% 23% 10% 10% 0% 28% Medium conversion 36% 19% 18% 0% 0% 26% Medium low water onshore Hydroskimming 29% 23% 10% 10% 0% 28% Medium conversion 36% 19% 18% 0% 0% 26% Medium low water offshore Hydroskimming 29% 23% 10% 10% 0% 28% Medium conversion 36% 19% 18% 0% 0% 26% Light watery onshore Hydroskimming 29% 23% 10% 10% 0% 28% Medium conversion 36% 19% 18% 0% 0% 26% Light watery offshore Hydroskimming 29% 23% 10% 10% 0% 28% Medium conversion 36% 19% 18% 0% 0% 26% Medium watery onshore Hydroskimming 29% 23% 10% 10% 0% 28% Medium conversion 36% 19% 18% 0% 0% 26% Medium watery offshore Hydroskimming 29% 23% 10% 10% 0% 28% Medium conversion 36% 19% 18% 0% 0% 26% Heavy low water onshore Medium conversion 36% 19% 18% 0% 0% 26% Deep conversion - coking 44% 11% 31% 0% 10% 3% Deep conversion - hydrocracking 38% 11% 40% 0% 8% 3% Heavy low water offshore Medium conversion 36% 19% 18% 0% 0% 26% Deep conversion - coking 44% 11% 31% 0% 10% 3% Deep conversion - hydrocracking 38% 11% 40% 0% 8% 3% Heavy watery onshore Medium conversion 36% 19% 18% 0% 0% 26% Deep conversion - coking 44% 11% 31% 0% 10% 3% Deep conversion - hydrocracking 38% 11% 40% 0% 8% 3% Heavy watery offshore Medium conversion 36% 19% 18% 0% 0% 26% Deep conversion - coking 44% 11% 31% 0% 10% 3% Deep conversion - hydrocracking 38% 11% 40% 0% 8% 3% Ultra-Deep Hydroskimming 29% 23% 10% 10% 0% 28% Medium conversion 36% 19% 18% 0% 0% 26% Fracking Hydroskimming 29% 23% 10% 10% 0% 28% Light and medium depleted Hydroskimming 29% 23% 10% 10% 0% 28% Medium conversion 36% 19% 18% 0% 0% 26% Heavy depleted Medium conversion 36% 19% 18% 0% 0% 26% Deep conversion - coking 44% 11% 31% 0% 10% 3% Deep conversion - hydrocracking 38% 11% 40% 0% 8% 3% • Extraction data processing from the OPGEE dataset Each oil field in the OPGEE_v.1.1_draft_e dataset presents different energy consumption and GHG emissions, that are reported in the “Energy consumption” and “GHG emission” sheets. Data from Table 3–6 (“Energy consumption”) and total direct (Table 1, “GHG emission”) and indirect emissions (Table 2, “GHG emission”) have been used to construct the sixteen typologies of extraction processors. OPGEE data are reported in MMBtu/ day and gCO2eq/day, so for our scopes they all have been converted to MJ/bbl and kgCO2eq/bbl of oil, using daily capacity production of respective fields. (See Tables A1-A3) For each oil field, the following set of data has been collected, out of the whole information: • Natural gas (NG): on-site (net) consumption; • Diesel: imports (equal to the net consumption by OPGEE assumption) and indirect consumption associated with imports; • Electricity, on-site (net) consumption (entirely imported due to OPGEE assumption) and indirect consumption associated with imports; • GHG emissions: direct emissions due to the consumption of energy carriers on-site and indirect consumption associated to diesel and electricity imports; Table A4 Mix of oil products, in quantity and quality, in the current situation and in the four simulations. Simulations Gasoline Jet Fuel ULSD Fuel Oil Coke LHE Current situation 36% 19% 18% 1% 1% 24% Simulation 1 36% 19% 18% 1% 1% 24% Simulation 2 36% 19% 18% 1% 1% 24% Simulation 3 36% 18% 19% 1% 1% 24% Simulation 4 34% 19% 16% 3% 1% 24% 16 M. Manfroni et al. Applied Energy 298 (2021) 117210 • Production capacity per field, in order to make the conversions and to assess the magnitude of each typology in the production mix. These are the five main categories of energy carriers. Direct GHG emissions are summed up and not reported for specific associated energy carrier. The dataset contains also information about: • Water-to-Oil ratio; • Field depth and location (onshore – offshore); • Sulfur content. This information was used to construct the extraction (functional) typologies, as reported in Table A5. Table A5 Assumptions for identifying the taxonomy of extraction (functional) processors. Adapted from OCI – Oil Climate Index [28] Classification API crude category (API degrees) Heavy 0–21,99 Medium 21,99–31,99 Light >31,99 Sulfur content (S %wt) Sweet <0,5 Sour 0,5 Depth (m) Shallow 0–2133,6 Ultra-Deep >3352,8 Water-to-Oil ratio (bbl water/bbl oil) Low water 0–4 Watery >4 In order to generate the unitary description and calculate the benchmarks for consumption of energy carriers per extraction type, values from specific oil fields have been weighted averaged on the production capacity. Using a bottom-up approach, we generated unitary benchmarks of extraction types observing special instances (structural elements) of oil fields: ∑n i=1 EC i,k Pr i ECk = ∑ n i=1 Pr i ∑n i=1 GHGi,k Pr i GHGk = ∑n i=1 Pr i Where k = category of carriers (for instance, NG exports), n = number of oil fields of the same type, EC = energy carrier in MJ/bbl (natural and refinery gas, diesel, electricity), GHG = GHG emissions in kgCO2eq/bbl, Pr = production capacity per day (bbl/day). In this way, it is possible to define a general taxonomy based on intensive (unitary) processors that can be tailored to specific cases by simply introducing the size of the system under analysis. Note that the average is weighted on the extraction capacity (volume of barrels extracted per day). Therefore, bigger fields assume more importance than smaller ones inside the network representation. • Refining data processing from the PRELIM dataset Data for each refinery have been collected from the “Main input & output” sheet of PRELIM model v1.2. In the column “Energy use”, types of carriers reported are heat, steam, hydrogen (SMR, Steam Methane Reformer and CNR, Catalytic Naphtha Reformer), electricity, FCC (Fluid Catalytic Cracking) catalyst Regeneration and excess of RFG (Refinery Fuel Gas). For our scope, we defined the following categories and performed a mapping with PRELIM categories: • Natural gas and refinery fuel gas (NG + RFG); • Electricity; • Diesel; • Coke; • GHG emissions. Electricity matches straightly between the two sets of accounting. Heat, steam and hydrogen SMR sub-categories are summed up and the total goes to “NG + RFG”. Hydrogen via CNR is supposed to be entirely obtained from natural gas. FCC Cat. Regeneration is supposed to be done through coke burning. Diesel use in refineries is zero. The “Diesel” category is conserved for matching with extraction categories, but there is no direct diesel use in refining (Table A2). Excess of RFG is always zero due to PRELIM assumptions. Regarding GHG emissions, total emissions are directly reported under the “GHG emissions” category in our model. Refineries have been grouped according to their configuration into the four typologies of the selected taxonomy and benchmarks obtained using simple average calculation. No weighted average has been performed this time, since refinery capacity production is elastic and very dependent on external factors (market). 17 M. Manfroni et al. Applied Energy 298 (2021) 117210 The categories of oil products selected for our model are reported in the list below, and another mapping with PRELIM’s one has been done: • Gasoline; • Jet Fuel; • Diesel; • Fuel Oil; • Coke; Gasoline maps directly with PRELIM’s “Blended Gasoline”, Jet Fuel with Jet-A/AVTUR, Diesel with ULSD (Ultra Light Sulphur Diesel), Fuel Oil, Coke and Liquid Heavy Ends (LHE) are the same of PRELIM’s. Surplus of Refiney Fuel gas is always zero for PRELIM’s assumption. • Network scaling with Metabolic Processors The semantic framework adopted by relational analysis is based on the four Aristotelian causes: (i) material cause, the material input coming from nature – this is useful to identify external constraints; (ii) formal cause, the organizational structure (recorded in blueprints/know-how) and the set of cybernetic controls that can be stored as recorded information – this is useful to study the role of technology and know-how; (iii) efficient cause, the agents of change expressing the required functions – this is the semantically open part of the representation, because the same function can be expressed by different combinations of material and formal causes; and (iv) final cause, defining the purpose of the energy system for the embedding social system. The analysis of energy systems based on metabolic processors provides three important advantages: 1. The set of relations characterized by a processor avoids the simplifications typical of reductionism A processor integrates a set of non-equivalent descriptions of the observed system in the form of a data array. The same structure of the data array can be used to characterize the metabolic attributes of both structural and functional elements and of functional elements generated by a combination of lower-level functional elements, when moving the analysis to upper levels. The data sources used to describe the various elements across levels are non-reducible to each other in a common metric. Thus, establishing congruent relations between processors describing structural elements and functional elements across levels allows to generate a coherent representation across scales inside the metabolic pattern. Data about the flows “coming from” and “going to” the technosphere (flows having a technical and an economic relevance) can be used to check the performance in terms of technical coefficients, labor requirement, economic costs and revenues. Data about the flows “coming from” and “going to” the biosphere (water, primary sources on the supply side and emissions and pollutants on the sink side) can be used to assess environmental pressures and, when geore­ ferenced, to check potential environmental impacts. Combining structural and functional representations in a coherent analytical tool allows to track what is produced and consumed, how is produced and consumed, at what biophysical and environmental costs in relation to which societal purposes. Five different categories of inputs and outputs are represented in Fig. A4: τi = secondary flow inputs: circulating flow elements required by processes coming from the technosphere (processes under human control), e.g., electricity, fuels; θi = secondary fund inputs: fixed fund elements required by processes coming from the technosphere (processes under human control), e.g., hours of workers, power capacity; βi = primary supplies of inputs coming from processes outside human control (biosphere): availability of primary sources (either fund-flow or stock depletion), e.g., oil, gas; Fig. A4. Representation of the data array associated with a metabolic processor. 18 M. Manfroni et al. Applied Energy 298 (2021) 117210 γi = primary sinks of outputs provided by processes outside human control (biosphere): availability of primary sink capacity (either fund-flow or sink filling), e.g., GHG emissions, polluting water; σ = secondary flow outputs: products required from society determined in both quantity and quality, e.g., crude oil, refinery products. The relations over the quantities of these different flow and fund elements can be expressed using two non-equivalent representations: • Extensive values: the values observed for specific processes; • Unitary values: technical coefficients calculated as benchmarks of given functional or structural elements. 2. It is possible to move the representation given by processors across different levels of analysis and scales, maintaining coherence between the various quantitative assessments Different definitions of a functional unit, viewed as an effective combination of lower-level components, are illustrated in Fig. A5. On the lower-left corner, different functional units of extractions (e.g., different types of fields operating at level n-2) are combined together to generate a functional unit of “extraction” at the level n-1. On the upper left corner, we see a sequential pathway of processes (“extraction” →“transportation” →“refinery”), which can be considered as a functional unit “oil products production” at level n (on the right-upper corner). Each sequential pathway is composed by a series of functional units defined at level n-1 (blue double arrow).Fig. A6. Using this accounting framework, we can integrate non-equivalent descriptions referring to structural and functional elements: (i) the WHAT/ HOW can be characterized by looking at the structural aspects of the elements: the “WHAT” is the specific transformation and the “HOW” is a characterization based on benchmarks of the processes carried out, i.e. the material infrastructures and the material flows coming in and getting out associated with the necessary know-how and the observed technical coefficients (bottom-up analysis); and (ii) the WHAT/WHY can be defined by looking at the functional role played by the element inside the rest of the system, i.e. the purposes justifying the expression of the process in the first place (why society supports the reproduction of it) and the specific combination of element in a functional units (top-down analysis). 3. It is possible to generate contextualization for the performance of the oil sector and its higher and lower-level components The energy sector is a component interacting with the rest society, the oil sector is a component of the energy sector, and the compartment of extraction is just a component of the oil sector. Since complex socio-ecological systems are hierarchically structured [71,72], keeping a holistic picture of the relations over hierarchical elements inside the metabolic pattern expressed at different levels is vital. This helps to check the correctness of the representation given and to grasp the meaning of different components inside the system (the big picture) [73]. Fig. A5. Moving the assessment of the oil sector across different scales (non-equivalent definitions of functional units) and hierarchical levels. 19 M. Manfroni et al. Applied Energy 298 (2021) 117210 Fig. A6. Map contextualizing the role of structural and functional elements within the energy sector and within the rest of the society. References [19] Nieto J, Carpintero Ó, Miguel LJ. Less than 2◦ C ? An economic-environmental evaluation of the Paris Agreement. Ecol Econ 2018;146:69–84. https://doi.org/ 10.1016/j.ecolecon.2017.10.007. [1] Vitousek PM, Mooney HA, Lubchenco J, Melillo JM. Human domination of Earth’s [20] Höök M, Hirsch R, Aleklett K. Giant oil field decline rates and their influence on ecosystems. Urban Ecol An Int Perspect Interact Between Hum Nat 1997;277:3–13. world oil production. Energy Policy 2009;37(6):2262–72. https://doi.org/ https://doi.org/10.1007/978-0-387-73412-5_1. 10.1016/j.enpol.2009.02.020. [2] Rockström J, et al. A Safe Operating Space for Humanity. Nature 2009;461. [21] Farrell AE, Brandt AR. Risks of the oil transition. Environ Res Lett 2006;1(1). https://doi.org/10.1073/pnas.1014402108. https://doi.org/10.1088/1748-9326/1/1/014004. [3] Independent Group of Scientists appointed by the Secretary-General, Global [22] Gordon D. Understanding Unconventional Oil, 2012. [Online]. Available: Sustainable Development Report 2019: The Future is Now - Science for Achieving https://carnegieendowment.org/files/unconventional_oil.pdf. Sustainable Development, 2019. doi: 10.1016/j.aodf.2009.10.015. [23] Visual Capitalist. https://www.visualcapitalist.com/map-worlds-biggest-oil-disc [4] IMF, Mitigating climate change - growth - and distribution-friendly strategies, overies-since-1868/, 2019. . 2020. [24] Brandt AR. Oil depletion and the energy efficiency of oil production: The case of [5] BP, Statistical Review of World Energy, 2020. [Online]. Available: https://www. California. Sustainability 2011;3(10):1833–54. https://doi.org/10.3390/ bp.com/content/dam/bp/business-sites/en/global/corporate/pdfs/energy-eco su3101833. nomics/statistical-review/bp-stats-review-2020-full-report.pdf. [25] Masnadi MS, Brandt AR. Energetic productivity dynamics of global super-giant [6] International Energy Agency IEA, Energy Technology Perspectives 2020, 2020. oilfields. Energy Environ Sci 2017;10(6):1493–504. https://doi.org/10.1039/ [7] Sgouridis S, Csala D, Bardi U. The sower’s way: Quantifying the narrowing net- c7ee01031a. energy pathways to a global energy transition. Environ Res Lett 2016;11(9). [26] Masnadi MS, Brandt AR. Climate impacts of oil extraction increase significantly https://doi.org/10.1088/1748-9326/11/9/094009. with oilfield age. Nat Clim Chang 2017;7(8):551–6. https://doi.org/10.1038/ [8] Solé J, García-Olivares A, Turiel A, Ballabrera-Poy J. Renewable transitions and the NCLIMATE3347. net energy from oil liquids: A scenarios study. Renew Energy 2018;116:258–71. [27] Parra R, Bukkens SGF, Giampietro M. Exploration of the environmental https://doi.org/10.1016/j.renene.2017.09.035. implications of ageing conventional oil reserves with relational analysis. Sci Total [9] Di Felice LJ, Ripa M, Giampietro M. Deep decarbonisation from a biophysical Environ 2020;749:142371. https://doi.org/10.1016/j.scitotenv.2020.142371. perspective: GHG emissions of a renewable electricity transformation in the EU. [28] Carnegie Endowment for International Peace. Oil Climate Index, 2018. http://oci. Sustain 2018;10(10). https://doi.org/10.3390/su10103685. carnegieendowment.org/#. [10] Gibbs HK, et al. Carbon payback times for crop-based biofuel expansion in the [29] Rogelj J, et al. Mitigation Pathways Compatible with 1.5◦ C in the Context of tropics : the effects of changing yield and technology. Environ Res Lett 2008;3: Sustainable Development. In: Global Warming of 1.5◦ C. An IPCC Special Report on 034001. https://doi.org/10.1088/1748-9326/3/3/034001. the impacts of global warming of 1.5◦ C above pre-industrial levels and related [11] Unruh GC. Understanding carbon lock-in. Energy Policy 2000;28:817–30. global greenhouse gas emission pathw, IPCC Spec. Rep. Glob. Warm. 1.5 oC, p. [12] Janipour Z, De Nooij R, Scholten P, Huijbregts MAJ, De Coninck H. What are 82pp, 2018, [Online]. Available: https://www.ipcc.ch/site/assets/uploads/sites/ sources of carbon lock-in in energy-intensive industry ? A case study into Dutch 2/2019/02/SR15_Chapter2_Low_Res.pdf. chemicals production. Energy Res Soc Sci 2020;60:101320. https://doi.org/ [30] Brandt AR, Farrell AE. Scraping the bottom of the barrel: Greenhouse gas emission 10.1016/j.erss.2019.101320. consequences of a transition to low-quality and synthetic petroleum resources. [13] Smil V. Power Density: A Key to Understanding Energy Sources and Uses. MIT Clim Change 2007;84(3–4):241–63. https://doi.org/10.1007/s10584-007-9275-y. Press; 2015. [31] European Commission. A Clean Planet for all. A European long-term strategic [14] Smil V. Energy and Civilization: A History. MIT Press; 2017. vision for a prosperous, modern, competitive and climate neutral economy, 2018. [15] Tong D, et al. Committed emissions from existing energy infrastructure jeopardize [32] Brandt AR, Masnadi MS, Englander JG, Koomey J, Gordon D. Climate-wise choices 1.5◦ C climate target. Nature 2019;572(7769):373–7. https://doi.org/10.1038/ in a world of oil abundance. Environ Res Lett 2018;13(4). https://doi.org/ s41586-019-1364-3. 10.1088/1748-9326/aaae76. [16] Röck M, et al. Embodied GHG emissions of buildings – The hidden challenge for [33] Masnadi MS, et al. Global carbon intensity of crude oil production. Science 2018; effective climate change mitigation. Appl Energy 2020;258:114107. https://doi. 361(6405):851–3. https://doi.org/10.1126/science.aar6859. org/10.1016/j.apenergy.2019.114107. [34] Brandt AR, et al. Energy Intensity and Greenhouse Gas Emissions from Tight Oil [17] Davis SJ, Caldeira K, Matthews HD. Future CO2 Emissions and Climate Change Production in the Bakken Formation. Energy Fuels 2016;30(11):9613–21. https:// from Existing Energy Infrastructure. Science 2010;329(5997):1330–3. https://doi. doi.org/10.1021/acs.energyfuels.6b01907. org/10.1126/science.1188566. [35] Jing L, et al. Carbon intensity of global crude oil refining and mitigation potential. [18] UNEP, Emissions gap report 2019, 2019. Nat Clim Chang 2020;10(6):526–32. https://doi.org/10.1038/s41558-020-0775-3. 20 M. Manfroni et al. Applied Energy 298 (2021) 117210 [36] Farrell AE, Plevin RJ, Turner BT, Jones AD, O’Hare M, Kammen DM. Ethanol can [53] Ramírez-Corredores MM. The science and technology of unconventional oils : contribute to energy and environmental goals. Science 2006;311(5760):506–8. finding refining opportunities. Academic Press; 2017. https://doi.org/10.1126/science.1121416. [54] González-López R, Giampietro M. Multi-Scale Integrated Analysis of Charcoal [37] Wang M, Huo H, Arora S. Methods of dealing with co-products of biofuels in life- Production in Complex Social-Ecological Systems. Front Environ Sci 2017;5:1–16. cycle analysis and consequent results within the U.S. context. Energy Policy 2011; https://doi.org/10.3389/fenvs.2017.00054. 39(10):5726–36. https://doi.org/10.1016/j.enpol.2010.03.052. [55] M. Giampietro et al., Report on the Experience of Applications of the Nexus [38] Heijungs R, Suh S. The Computation Structure of Life Cycle Assessment. Kluwer Structuring Space in Quantitative Storytelling, 2020. Academic Publishers; 2002. [56] ENI. World Oil Review 2018, 2018. [39] IPCC. Fifth Assessment Report, 2014. [Online]. Available: Available from /http:// [57] Kapustin NO, Grushevenko DA. Global prospects of unconventional oil in the www.ipcc.ch/. turbulent market: A long term outlook to 2040. Oil Gas Sci Technol 2018;73:1–13. [40] Rissman J, et al. Technologies and policies to decarbonize global industry: Review https://doi.org/10.2516/ogst/2018063. and assessment of mitigation drivers through 2070. Appl Energy 2020;266: [58] Velasco-Fern R, Giampietro M, Bukkens SGF. Analyzing the energy performance of 114848. https://doi.org/10.1016/j.apenergy.2020.114848. manufacturing across levels using the end-use matrix. Energy 2018;161. https:// [41] Zappa W, Junginger M, van den Broek M. Is a 100% renewable European power doi.org/10.1016/j.energy.2018.07.122. system feasible by 2050? Appl Energy 2019;233–234:1027–50. https://doi.org/ [59] Hongjun W, et al. Assessment of global unconventional oil and gas resources. Pet 10.1016/j.apenergy.2018.08.109. Explor Dev 2016;43(6):925–40. https://doi.org/10.1016/S1876-3804(16)30111- [42] Bloess A, Schill WP, Zerrahn A. Power-to-heat for renewable energy integration: A 2. review of technologies, modeling approaches, and flexibility potentials. Appl [60] Gordon D, Mathews JT. Smart Tax: Pricing Oil for A Safe Climate, 2016. Energy 2018;212:1611–26. https://doi.org/10.1016/j.apenergy.2017.12.073. [61] Allison E, Mandler B. Petroleum and the Environment. The Am Geosci Inst 2018. [43] Scrieciu S, Rezai A, Mechler R. On the economic foundations of green growth [62] Craig Jr FF. The Reservoir Engineering Aspects of Waterflooding. SPE - Soc discourses: The case of climate change mitigation and macroeconomic dynamics in Petroleum Eng 1971. economic modeling. Wiley Interdiscip Rev Energy Environ 2013;2(3):251–68. [63] Haige W, Yunhua G, Lin S. Technologies in deep and ultra-deep well drilling: https://doi.org/10.1002/wene.57. Present status, challenges and future trend in the 13th Five-Year Plan period [44] Rosen RA, Guenther E. The economics of mitigating climate change: What can we (2016–2020). Nat Gas Ind B 2020;4(5):319–26. https://doi.org/10.1016/j. know? Technol Forecast Soc Change 2015;91:93–106. https://doi.org/10.1016/j. ngib.2017.09.001. techfore.2014.01.013. [64] Brandt AR, Yeskoo T, Vafi K. Net energy analysis of Bakken crude oil production [45] Höök M, Tang X. Depletion of fossil fuels and anthropogenic climate change - A using a well-level engineering-based model. Energy 2015;93:2191–8. https://doi. review. Energy Policy 2013;52:797–809. https://doi.org/10.1016/j. org/10.1016/j.energy.2015.10.113. enpol.2012.10.046. [65] Climate Watch Data, https://www.climatewatchdata.org/, 2016. [46] Capellán-Pérez I, Arto I, Polanco-Martínez JM, González-Eguino M, Neumann MB. [66] Maugeri L, Oil: the next revolution, 2012. Likelihood of climate change pathways under uncertainty on fossil fuel resource [67] McNally MS, Brandt AR. The productivity and potential future recovery of the availability. Energy Environ Sci 2016;9(8):2482–96. https://doi.org/10.1039/ Bakken formation of North Dakota. J Unconv Oil Gas Resour 2015;11:11–8. c6ee01008c. https://doi.org/10.1016/j.juogr.2015.04.002. [47] Brecha RJ. Emission scenarios in the face of fossil-fuel peaking. Energy Policy [68] Beltrao RLC, Sombra CL, Lage ACVM, Netto JRF, Henriques CCD. Pre-salt Santos 2008;36(9):3492–504. https://doi.org/10.1016/j.enpol.2008.05.023. basin - Challenges and New Technologies for the Development of the Pre-salt [48] Berg P, Boland A. Analysis of Ultimate Fossil Fuel Reserves and Associated CO2 Cluster, Santos Basin, Brazil. Proc Offshore Technol Conf 2009. https://doi.org/ Emissions in IPCC Scenarios. Nat Resour Res 2014;23(1):141–58. https://doi.org/ 10.4043/19880-ms. 10.1007/s11053-013-9207-7. [69] Diaz-Maurin F. Power capacity: A key element in sustainability assessment. Ecol [49] IEA. World Energy Outlook 2019, 2019. Indic 2016;66:467–80. https://doi.org/10.1016/j.ecolind.2016.01.044. [50] González-López R, Giampietro M. Relational analysis of the oil and gas sector of [70] Laugeri L. The Age of Oil. The Mythology, History, and Future of the World’s Most Mexico: Implications for Mexico’s energy reform. Energy 2018;154:403–14. Controversial Resource. Praeger 2006. https://doi.org/10.1016/j.energy.2018.04.134. [71] Allen TFH, Starr TB. Hierarchy: Perspective for Ecological Complexity. University [51] Rosen R. Life Itself: A Comprehensive Inquiry Into the Nature, Origin, and of Chicago Press; 1982. Fabrication of Life. New York: Columbia University Press; 2005. [72] Pattee HH. Hierarchy theory; the challenge of complex systems. G. Braziller, 1973. [52] Louie AH. Intangible Life: Functional Connections in Relational Biology. Cham, [73] Giampietro M, Mayumi K, Sorman A. The Metabolic Pattern of Societies: Where CH: Springer International (Anticipation Series); 2017. Economists Fall Short. New York: Routledge; 2012. 21