Energy 290 (2024) 130264 Contents lists available at ScienceDirect Energy journal homepage: www.elsevier.com/locate/energy The many faces of heating transitions. Deeper understandings of future systems in Sweden and beyond Nick Martin a, *, Jakob Zinck Thellufsen b, Miguel Chang b, c, Laura Talens-Peiró a, Cristina Madrid-López a a Sostenipra Research Group (2021SGR 00734), Institute of Environmental Science and Technology (ICTA-UAB) (Maria de Maeztu MDM CEX2019-0 0 0940-M Excellence Unit), Autonomous University of Barcelona, 08193 Cerdanyola Del Vallès, Barcelona, Spain b Department of Planning, Aalborg University, Rendsburggade 14, 9000 Aalborg, Denmark c Department of Energy Systems Analysis, Institute for Energy Technology (IFE), 2027 Kjeller, Norway A R T I C L E I N F O A B S T R A C T Handling Editor: Dr. Henrik Lund Swedish buildings require additional heating inputs for many months each year, much of which comes from district heating systems fed by fuel combustion, electrical devices and recycled heat sources; localised heating is Keywords: also generated within individual buildings. Optimised projections using the EnergyPLAN model for a so-called Energy transition “smart energy” scenario predict dramatic reductions in biomass use accompanied by increases in electricity, District heating recycled heat and biogas. Electricity generation routes are also predicted to change, shifting away from nuclear Energy modelling and biomass sources towards wind, solar and biogas technologies. While such transitions are expected to lower Renewable energy Life cycle assessment greenhouse gas emissions, current assessment methods rarely examine the breadth of other environmental and Critical raw materials material supply aspects at play. Here, a novel new approach provides deeper insights into current and future Swedish heating scenarios for 11 key indicators by considering heat production processes over their full life cycles. Results suggest that favourable reductions are likely in five of these indicators, but these benefits are offset by unfavourable outcomes in others, including all three raw material supply indicators. Ultimately, the study provides a novel example of ways in which additional tools can complement existing modelling techniques and expand the scope of available information used to assist heating transition decisions. 1. Introduction rise in popularity of district heating as most new buildings were designed specifically to include district heating connectivity [4]. As with all Nordic countries, Sweden has a cold climate and heating Nevertheless, coal was the dominant fuel in these district heating sys­ is required for eight to 10 months of the year in most regions [1]. As a tems until the end of the 1980s when biomass began to take over the consequence, it consumes approximately 50 % more heat energy per market [5]. person, on average, than the EU as a whole. Not surprisingly, the use of Outside of centralised district heating systems, deregulation of the district heating in Sweden is high: in 2014 approximately 93 % of electricity market in 1996 led to an increase in the use of individual heat multi-residence buildings were connected to a district heating network pumps [4], which have continued to gain acceptance in the last 20 years [2]. However, this has not always been the case. to become the main competitor to district heating overall [2]. In fact, Fig. 1 displays the market share percentages of the different heating Sweden is now one of the top countries in the world for heat pump techniques employed to heat buildings in Sweden between 1960 and ownership per capita [1] and over half of its residential buildings now 2020. The data indicates that the use of individual oil-based heaters have at least one heat pump installed. In this arena, small-sized units dominated the market during the 1960s, but sharply declined in popu­ tend to be installed in one and two dwelling houses and medium-sized larity in the wake of the global oil crisis in the 1970s [3]. District heating units are used in apartment blocks and commercial buildings; much subsequently rose in popularity and has continued to steadily increase larger units are used in district heating applications. Lastly, many resi­ its market share, mostly under municipal ownership. A national pro­ dents–particularly those in rural and isolated areas of the country where gram to develop public housing between 1965 and 1974 assisted in the district heating systems are not present–still rely on the traditional * Corresponding author. E-mail address: nicholas.martin@uab.cat (N. Martin). https://doi.org/10.1016/j.energy.2024.130264 Received 29 April 2023; Received in revised form 2 January 2024; Accepted 3 January 2024 Available online 4 January 2024 0360-5442/© 2024 The Authors. 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/). N. Martin et al. Energy 290 (2024) 130264 method of heating their homes via the combustion of oil, biomass and pumps, boilers and heaters–is widely seen as a way of reducing the other fuels. Indeed, Fig. 1 suggests that this method of heating still oc­ environmental burdens associated with heating systems, especially in cupies around 9 % of the heating system overall. relation to greenhouse gas (GHG) emissions. Unfortunately, the majority District heating is a method for distributing thermal energy–in the of existing studies that incorporate the environmental impacts associ­ form of steam or hot water–throughout a series of insulated pipes to ated with heating systems tend to focus solely on GHG emissions. provide heating to multiple buildings within a local network [7]. It is Additionally, within this limited scope, they tend to only consider mostly used for space heating applications and to provide hot water to emissions that occur during the final combustion processes that produce residents and businesses. At present, heat used in district heating sys­ heat energy within these systems. In doing so, these studies neglect the tems is predominantly derived from the combustion of fuels [8]. This vast number of other emissions–and, indeed, environmental bur­ can be undertaken in heat-only boilers and alongside electricity gener­ dens–that occur in other parts of a heating system. Furthermore, they ation processes in combined heat and power (CHP) plants. Heat can also ignore the various material extraction, manufacturing, transportation, be obtained from various forms of recycled heat; the use of excess heat maintenance and disposal stages involved in creating and operating from industrial processes and the combustion of fuels condensed from heating infrastructure. In an attempt to fill this gap, many are now flue gases are common examples of this. Thermal energy is sometimes suggesting that transition pathways should be scrutinised with more also transferred directly to district heating networks from geothermal rigour so that a deeper and more robust set of environmental and other and solar sources. Lastly, electricity can be used to generate system heat indicators [12,13] can be identified. Doing so would allow analyses to using electric boilers and heat pumps of different sizes. look beyond simple economic and GHG targets and to consider the im­ Much like traditional electric boilers and heaters, heat pumps are pacts relating to all stages and physical aspects of heat production devices that use electrical energy to produce volumes of warm air or processes [14]. water suitable for heating applications. In a heat pump system, a More recently, the incorporation of life cycle assessment (LCA) compressor is used to circulate a refrigerant within a closed loop in order methodologies into the analysis and modelling of energy systems has to amplify temperature differentials, much like an air conditioner in been specifically highlighted as a way of integrating these improvements reverse [1]. Here, the coefficient of performance (COP) reflects the ratio [15–18] and a wide range of studies are now beginning to use LCA of heat energy produced to electrical energy used; values of COP depend concepts for examining transition strategies. The majority of past energy on the temperature differentials at play but are generally between two related LCAs have investigated specific technologies or locations [19], and seven for most applications [9,10]. As such, heat pumps offer effi­ although recent efforts have attempted to examine outputs for entire cient and attractive pathways for heating spaces using electricity, systems, typically by pairing LCA calculations with the outputs of energy particularly if renewable forms of electricity can be used. As a result, the models. For example, a significant amount of progress has been made to use of heat pumps is projected to become an important element of sus­ link global integrated assessment models (IAMs) with LCA calculations tainable heating systems and they are widely predicted to achieve wider [16,20–22] and to smaller national or regional energy models [14,17, presence in the district heating systems of the future [4]. More impor­ 23]. It is also worth noting that the usefulness of LCA data in assessing tantly, they are also rapidly gaining popularity as a standalone method energy systems need not be limited to the direct use of impact indicators. for vastly improving the efficiency of heating in buildings [11], partic­ Inventory data from LCA databases can also be used to provide raw ularly in areas where district systems are not available or viable. material data for processes on a per-unit basis. As such, this data can also Transitioning away from the combustion of fossil fuels in producing be used to derive a variety of material-related indicators using cus­ heat energy–and, indeed, in producing the electricity used to power heat tomised methodologies. Fig. 1. Generation sources, by percentage of energy share, for heat supply to residential and service sector buildings within the Swedish energy system. Note that all forms of electrical heat generation are grouped together, including those used in district heating networks. As such, district heating totals do not include the electrical component. The “Oil” and “Other” categories incorporate all other sources at the localised building level. This includes, for example, the combustion of oil, logs, wood pellets and natural gas in decentralised boilers, fireplaces and furnaces. Data sources [2,6]. 2 N. Martin et al. Energy 290 (2024) 130264 In any case, Bartolozzi et al. [24] report that, historically, very few To fill this gap, a methodology was derived for undertaking more LCA studies have been conducted that specifically evaluate district detailed environmental assessments of full district heating configura­ heating systems. Furthermore, the identified studies–and, indeed, the tions. To the best of our knowledge, the study represents the first time few more recent investigations–are almost all centred upon specific that the LCA approach has been applied to a national level system and technologies or stages of a process, or on systems at the facility or city incorporates all aspects of a heating system, including the electricity scale [25–27]; similar conclusions about the scarcity of detailed inputs to a system and the contributions of both district heating and LCA-based analyses were also drawn by Jeandaux et al. [28]. isolated heat generation. Moreover, it is one of the few studies designed Fig. 2. Representation of Swedish heating system used in the analysis. District heating systems receive inputs from the direct combustion of fuels, heat generated from utility-scale electrical heat pumps and boilers, and from recycled heat sources. Locally generated forms of heat are also included, assumed to be from smaller- scale electrical boilers, heaters and heat pumps, and from combustion boilers, fireplaces and furnaces assumed to predominantly burn wood and oil, all of which occurs at the building level. All electrical inputs to both groups are disaggregated into typical technological categorisations and scaled according to “energy mix” data for electricity within a given scenario. Note that heat and electricity from waste and recycled heat sources–shown shaded in the figure–are not included in the impact analyses as they are assumed to be essentially “neutral” processes. However, in reality, some smaller impacts are still likely to be generated in relation to the infrastructure related to these activities. 3 N. Martin et al. Energy 290 (2024) 130264 to facilitate a thorough analysis and comparison of historical system 2.2. Additional specifications configurations and future configurations defined by optimised model­ ling outputs. With the system specified, a selection of additional specification data Here, we apply a newly derived approach [14]–developed within the is required to define the way in which the analysis is undertaken. First of Sustainable Energy Transitions Laboratory (SENTINEL) project [29]–to all, individual life cycle inventory (LCI) data is required for each of the enable a wide variety of environmental and material supply indicators to structural processors in Fig. 2; these are taken from v3.8 of the Ecoinvent be quantified using historical data for the Swedish heating system. More LCA database [36,37]. Where possible, specific processes for Sweden are importantly, we couple the workflow to outputs from the EnergyPLAN chosen. Where this is not possible, nearby countries or regional pro­ model [30,31] to provide a more-detailed overview of a potential future cesses for Europe are used; global or “rest-of-world” values are used only scenario for the system. In doing so, we provide a further demonstration when no other appropriate processes are available. A summary of the of the broader range of useful indicators that can be produced when processes assigned to each processor in the defined system is given in linking energy models with methodologies that allow more complete Table S1 in the supplementary material. Where multiple suitable listings assessments of environmental outcomes to be made using detailed LCA are available, an effort is made to select processes that represent the data. average or most typical values in that category to avoid biasing issues The analysis begins by defining the system in accordance with the resulting from processes that reflect outlier values. Again, as no in­ available data, including all inputs to district heating networks and dicators are to be generated for waste incineration, no definitions are locally generated heat sources. Separate analyses are then undertaken supplied for these processes. for historical configurations from 2015 and 2019 and compared to a Final indicator values are then able to be calculated using the stan­ projected “smart energy” scenario for the system in 2050 using results dard workflow [14]. The majority of these indicators use life cycle from the EnergyPLAN model. The sections that follow provide a thor­ impact assessment (LCIA) methods to generate a range of environmental ough description of the input data, presentation of the results obtained impact and resource use indicators. These methods are again taken from for a range of indicators and a final discussion of the key outcomes. v3.8 of the Ecoinvent LCA database; all selections are part of the “ReCiPe Midpoint (H)” group [38]. Values for three additional raw material in­ 2. Methodology dicators are also derived using the material requirement values from LCI listings in conjunction with previously defined methodologies [39,40]. In order to assess the characteristics of different historical and future A summary of the methods adopted for calculating the final 12 in­ configurations of the Swedish heating system, a customised version of dicators is provided in Table 1. the ENBIOS workflow [14] is defined and implemented, as described in the sections that follow. 2.3. Historical data 2.1. System definition The Swedish Energy Agency (Energimyndigheten) provides detailed annual data for district heating and electricity supply systems as part of The first stage of the analysis involves the definition of a customised its “Electricity supply, district heating and supply of natural gas” (“El-, hierarchical structure for the system at hand. The structure is based on gas-och fjärrvärmeförsörjningen”) series of reports. For district heating, the “dendrogram” concept defined within the MuSIASEM accounting the reports provide totals for the generation of district heating from method [32] and is governed by the available data and goals of the fuels, and the breakdown of which fuel inputs provided these outputs for analysis. In this case, a dendrogram was created to represent the supply both CHP plants and heat-only boilers. Accordingly, total amounts of of heat energy within the Swedish energy system, as shown in Fig. 2. energy generated from each fuel and plant type can be calculated. Here, Here, the hierarchy captured in the dendrogram is defined in accordance ten of the most common fuel/plant combinations are used; where only with the available categories provided in the historical data [33,34] and one plant type is significant, the combined totals of both are used. Values projected modelling outputs [35] for Sweden. Firstly, it includes the are also provided for industrial excess heat and flue gas condensation, heat energy distributed within district heating systems. This includes collectively assigned as “recycled heat” sources. Calculated final totals heat derived from the direct combustion of fuels, from utility-scale using the 2015 [41] and 2019 [33] versions of the report are shown in electrical infrastructure such as heat pumps and boilers, and from so Table 2. Values are also shown for heat created via local generation called “recycled” sources like industrial excess heat and flue gas processes–i.e., those outside of district heating systems–based on data condensation. Secondly, heating generated in individual buildings from the Swedish Energy Agency and available elsewhere [2,6]. It is outside of district heating systems are considered. This includes the worth noting that values for 2020 are not used in the analysis as they are combustion of oil, biomass and other fuels in local boilers, fireplaces and significantly different than those in preceding years as a result of drastic furnaces and the use of smaller-scale electrical heat pumps, boilers and changes in energy use patterns resulting from the COVID-19 pandemic. heaters at the building level. Electricity inputs into the electric boilers and heat pumps used In the end, 27 structural processors at the “n-5” level are aggregated within district heating systems for these years are also listed in the into 10 source categories at the “n-4” level: hydro, wind, solar, wood/ respective reports and are summarised in Table S2 in the supplementary biomass, waste incineration, biogas, natural gas, coal, oil and nuclear. material. No specific historical data is available for the use of electricity These are further grouped into three renewable energy classes at the “n- in heat pumps, heaters and boilers outside of district heating systems. 3” level: renewable, non-renewable, and bioenergy and waste. The However, the total amounts of electricity used to generate heat in all direct use of fuels and the use of electricity are then delineated at the “n- applications are also available elsewhere [2,6]. Subtracting district 2” level; the utilisation of recycled heat is also included at this level. heating input amounts from these totals, therefore, provides the totals Ultimately, the heating system is reduced to district heating and local used to generate heat from electricity at local levels. The heat totals generation processes at the “n-1” level. It should be noted that heat and created in these processes are then calculated assuming direct 1:1 con­ electricity generated from the incineration of waste and recycled heat version for boilers and heaters and a COP of 2.0 for heat pumps, as per sources are included in the system definition but are not considered in previous historical estimates for Sweden [6]; it is noted that historical the impact analyses beyond the summary of energy sources. This is COP values for district heating are reported to be between 3.9 and 4.2. because such sources are considered to be essentially “neutral” processes Thorough breakdowns of electricity generation are also provided by based on the use of existing sources of energy that would otherwise go the Swedish Energy Agency for 2015 [41] and 2019 [33]. A summary of unutilised. Nevertheless, it is recognised that, in reality, some impacts the electricity mix, by technology, is shown in Table 3. Note that no would be generated from the infrastructure related to these processes. distinction between onshore and offshore wind energy is provided in the 4 N. Martin et al. Energy 290 (2024) 130264 Table 1 Listing of methodologies used in deriving final indicators. A full listing is contained in Table S1 in the supplementary material. Group Indicator Method Units Total energy Energy generation Summing heat output values TWh LCIA [37] GHG emissions “climate change”, GWP100 Tg CO₂-eq Land occupation “agricultural land occupation”, ALOP + “urban land occupation”, ULOP x103 km2 Water depletion “water depletion”, WDP TL Fossil depletion “fossil depletion”, FDP Tg oil-eq Metal depletion “metal depletion”, MDP Tg Fe-eq Freshwater eutrophication “freshwater eutrophication”, FEP Gg P-eq Marine eutrophication “marine eutrophication”, MEP Gg N-eq Human toxicity “human toxicity”, HTPinf Tg 1-4-DC Raw materials Material supply risk [39,40] yr Env impacts relating to material supply [39] yr Env justice issues relating to material supply [39] yr 2.4. Projected data Table 2 Summary of historical and projected heat generation in district heating systems Predictions for 2050 are provided by modelled outputs from the and at the local generation level. A single plant type is use in instances where one type dominates the observed data or where only one type is represented in the EnergyPLAN model [30,31], as listed in Table 2. In particular, results LCI database. Sources [2,6,33,35,41]: were provided for Sweden [35] as a sub-region within a wider model simulation of a “smart energy” scenario [43,44] for the European energy Fuel Plant type Heat generation system created as part of the sEEnergies project [45]. This scenario seeks 2015 2019 2050 to maximise the use of renewable energy technologies while creating a [GWh] [GWh] [GWh] more flexible energy system where interactions between sectors are District heating optimised. Outputs from the model provide a thorough inventory of Recycled heat 8816 9377 18,190 district heating outputs from boilers and CHP units for a range of fuels. Fuels Wood chips CHP 12,066 12,188 1,939 However, in this scenario it was assumed that all energy produced in Boiler 5,622 5,671 463 boilers and CHP units are fed by wood chips, wood pellets and biogas; Wood pellets Boiler 2,637 2,260 184 Waste CHP 12,019 13,893 5,876 natural gas and all other fossil fuels are assumed to have been phased out Biogas CHP 30 116 7,754 by 2050. Nonetheless, it is noted that EnergyPLAN provides a single Natural gas CHP 1,221 741 total for biomass use and no distinction is made between raw wood chips Boiler 85 23 and processed pellets. Therefore, to maintain this delineation and allow Coal CHP 1,938 1,270 comparison with the historical scenarios, the biomass total is split ac­ Oil CHP 371 348 Boiler 397 251 cording to the observed ratio for 2019 [33]. Values for recycled heat Electricity 4,503 3,899 14,992 sources are also given, as are the totals from the local combustion of Local generation fuels, assumed to be entirely derived from wood products in 2050. Electricity 24,989 29,329 31,338 The amounts of heat generated from electricity in the 2050 scenar­ Fuels Wood 8,400 7,210 4,250 io–in district heating and localised applications–are also shown in Oil 900 600 Table 2; a full listing of the inputs and outputs for different infrastruc­ ture at each of these levels is also provided in Table S2 in the supple­ mentary material. Note that the values for electricity inputs to Table 3 electrolysers–that generate hydrogen later used in district heating net­ Summary of historical and projected electricity mix by technology. Sources [33, 35,41]: works–and for heat supplied from solar thermal infrastructure are also supplied in the EnergyPLAN data. Here, these values are arbitrarily Technology group Plant type Share in electricity mix bundled into the electricity totals for district heating because no life 2015 2019 2050 cycle assessment (LCA) data is available for these processes. [%] [%] [%] Again, electricity inputs and heat output totals from heat pumps are defined in accordance with a given COP value; a value of 4.0 was Hydro 46.6 38.8 21.9 Wind Onshore 10.0 11.8 47.7 assumed in the EnergyPLAN calculations for district heating. However, Offshore 0.004 16.5 for localised heat pump use, only the electricity inputs to these devices Solar 0.1 0.4 7.5 are included in the EnergyPLAN results. Although a conservative COP Wood CHP 5.5 6.6 0.6 value of 2.0 was used for historical heat pump data calculations [6], a Conventional 0.002 0.002 0.8 Waste incineration 1.7 2.1 0.9 COP of 3.4 was assumed when calculating outputs from future local heat Biogas 0.005 0.008 4.1 pumps, in line with estimates for more advanced household heat pump Natural gas CHP 0.3 0.2 technologies [46]. This also enables the total heating supply to be Conventional 0.04 approximately in line with historical totals. Meanwhile, all boilers are Coal CHP 0.6 0.5 again assumed to simply convert electricity to heat in a 1:1 ratio. Conventional 0.2 0.1 Oil CHP 0.2 0.2 For the electricity itself, EnergyPLAN provides a detailed breakdown Conventional 0.1 0.04 of all contributing technologies, including the split between CHP and Nuclear 34.8 39.3 boilers, where appropriate. A summary of the projected electricity mix, by technology, is listed in Table 3. The sum of electrical inputs to heat generation can then be proportioned pro-rata to the different electrical data. Despite this, the Global Wind Energy Council (GWEC) provides generation processes. breakdowns of installed capacity for Sweden for both years [42] and this A breakdown of the total inputs to the Swedish heating system–as data was used to further delineate the single wind energy totals. defined in Fig. 2–is provided in Fig. 3. It includes historical data for 2019 Nevertheless, the totals for offshore wind remain very low. [6,33] and projected values for 2050 [35]. The data indicates that the 5 N. Martin et al. Energy 290 (2024) 130264 Fig. 3. Historical and projected percentage breakdowns of total heat generation. A further breakdown of the electricity component is given in Fig. 4. Data sources [6, 33,35]. share of electricity use is expected to rise from 38 % to 55 %. Most supplementary material–are used in conjunction with the technological notably, this projection assumes a drop in wood biomass use from 31 % mix data in Table 3. to around 8.0 %. The direct use of waste in incinerators is predicted to decrease from 16 % to 6.9 %, while biogas derived from waste will rise 3.1. Summary of overall changes from negligible levels in 2019 to around 9.1 % in 2050. All fossil fuels will be eliminated. The breakdown of electricity generation, as illus­ A summary of the percentage changes forecast to occur between trated in Fig. 4, is also expected to change. The biggest change is the 2019 and 2050 is given in Table 4. Firstly, as a sanity check, the total complete elimination of nuclear power, which represented 39 % of amount of heat energy generated in the system under each scenario was generation in 2019. Wind energy is the biggest mover in replacing nu­ assessed and was only observed to vary by 2.5 %; this is well within the clear power, rising from 12 % to 64 %; this move also impacts hydro range of typical annual variations. Of the 11 proper indicators, adverse power, which drops from 39 % to 22 %. Solar power also makes a changes–where increases are observed–are predicted to occur in six in­ noticeable impact in the electricity market, rising from negligible levels dicators. All of these changes are notably high, from a 98.7 % increase in in 2019 to attain a 7.5 % share by 2050. Fossil fuels are also eliminated freshwater eutrophication to a rise of 329.6 % in relation to the envi­ in the electricity sector and the use of wood and waste reduces signifi­ ronmental impacts generated from raw material extraction. Beneficial cantly. Biogas is again predicted to achieve a small penetration in the changes between 2019 and 2050 are observed in the remaining five market, achieving a 4.1 % share by 2050. indicators, all of which reduce by at least 38 %. The phasing out of fossil fuels, wood biomass and nuclear power by 2050 has resulted in many 3. Results of analysis reductions in key indicators. However, replacing these sources with other renewable energy and bioenergy technologies is also shown to With the system specified and all historical and projected data in have detrimental effects on future outcomes, especially with respect to place, results are generated at each processor in the dendrogram for each wind, solar and biogas sources. of the indicators listed in Table 1. Scaling of LCI processes is performed using the final amounts of energy that relate to each processor. For heat, 3.2. Findings for key indicators the final generation values shown in Table 2 are used. For electricity, the total amount of energy input values–as listed in Table S2 in the Results for six key indicators–total heat generation, GHG emissions, Fig. 4. Historical and projected percentage breakdowns of electricity generation. Data sources [33,35]. 6 N. Martin et al. Energy 290 (2024) 130264 Table 4 Summary of percentage changes observed for 12 indicators between 2019 and 2050. Potentially adverse results are displayed in shaded cells. A summary of the key determinants of the predicted changes is also provided. Group Indicator Observed change (2019–2050) Key determinants of changes Total energy Total heat generation − 2.5 % Annual variations LCIA GHG emissions − 38.0 % Phasing out of coal, oil & wood Land occupation − 64.9 % Phasing out of wood Water depletion +251.7 % Phasing out of nuclear Replaced by biogas Fossil depletion − 49.6 % Phasing out of coal, oil, natural gas & wood Metal depletion +110.8 % Wind, solar & biogas Freshwater eutrophication +98.7 % Phasing out of coal & wood Replaced by biogas & wind Marine eutrophication − 45.3 % Phasing out of wood Human toxicity − 47.5 % Phasing out of wood Raw materials Material supply risk +107.2 % Wind, solar & biogas Env impacts relating to material supply +329.6 % Wind, solar & biogas Env justice issues relating to material supply +139.3 % Wind, solar & biogas land occupation, water depletion, material supply risk and environ­ model. At the “n-2” level, both fuel-based approaches are observed to mental impacts relating to material supply–are displayed in individual lose ground to electrical approaches, particularly in district heating figures in the sections that follow. Results for the remaining six in­ systems. Likewise, the use of inputs from recycled heat sources is seen to dicators are provided in section S3 in the supplementary material. Each almost double between 2019 and 2050. The data at the “n-3” and “n-4” row of figures illustrates the totals for a given indicator for the two levels–which do not show recycled heat–confirm the projected move historical system configurations–2015 and 2019–alongside the pro­ away from fossil fuel and nuclear technologies towards cleaner options. jected system for 2050. Representations are shown for three hierarchical In fact, the generation of electricity and heat from wood and waste levels: “n-2” for fuel combustion and electricity, “n-3” for renewability combustion are also seen to drop away notably. In their place, the use of category, and “n-4” for individual technology categories. Brief discus­ biogas in CHP plants is predicted to rise significantly. And, while solar sions are also provided for each indicator. energy is expected to make inroads into the electricity market, by far the biggest shift at the “n-4” technological level is in the increased use of 3.2.1. Total heat generation wind energy. Overall heat energy generation totals–which include all heat pro­ duction processes and recycled sources–are predicted to remain rela­ 3.2.2. GHG emissions tively constant across the three years presented, as shown in Fig. 5. Perhaps the most policy-relevant indicator presented here is that of Indeed, the total in the modelled system for 2050 is within 2.5 % of the total GHG emissions. Here, following a dramatic decrease of 17.0 % total recorded in 2019. Observing the historical values of this indicator between 2015 and 2019, GHG emissions drop a further 38.0 % by 2050, between 2015 and 2019 reveals that changes of between 1 % and 8 % as shown in Fig. 6. Results at the “n-3” level reveal that these reductions are commonplace from year to year [33,41,47–49]. This is hardly sur­ are largely linked to the non-renewable forms of heat. Further analysis at prising considering that milder or colder winters can easily affect overall the “n-4” level reveals the direct connection to coal and oil between heating requirements. As such, the similarities in total between 2019 2015 and 2019. The complete removal of these technologies, alongside and 2050 confirm that the system data has been suitably aligned and natural gas and nuclear, is a predictably dominant factor in the observed prepared, and that direct comparisons can be made between the recor­ changes by 2050; large reductions in wood use are also a factor. Despite ded and modelled systems for the remaining indicators. the large rises in wind and solar use predicted by 2050, these processes The data presented in Fig. 5 also provides a useful overview of the do not contribute significant amounts of GHG emissions. However, some fundamental changes that occur in the system between the recent his­ of these reductions are offset by the emissions created via the increasing torical configurations and the future configuration forecast by the use of derived biogas. Fig. 5. Comparison of historical and predicted values for total energy generation. Results are shown across three separate hierarchical levels. Note that recycled heat is not shown in the data at the “n-3” and “n-4” levels as it is fundamentally outside of the scope of these classifications. 7 N. Martin et al. Energy 290 (2024) 130264 Fig. 6. Comparison of historical and predicted values for GHG emissions. Results are shown across three separate hierarchical levels. Nevertheless, although no specific GHG emissions targets have been requirement, highlighting the dangers posed by bioenergy technologies stated for heating by the Swedish government, the country’s national in this regard. energy and climate plan [50] states that net GHG emissions must be reduced to zero and all electricity must be from renewable sources by 3.2.4. Water depletion 2040. As a result, the projected scenario from EnergyPLAN certainly Conversely, the required amounts of water are predicted to more does not come close to satisfying current policy targets using the as­ than triple between 2019 and 2050, rising by around 251.7 %, as shown sumptions used in this assessment. This, again, highlights the differences in Fig. 8. Simple visual inspection at the “n-3” and “n-4” levels imme­ between the assumptions made in LCA processes against those made in diately reveals that this rise is strongly linked to biogas production, other policy-based quantifications. It also highlights the importance of replacing the previous dominance of nuclear power which–according to changing “background” systems when making assessments for future the values presented in Table S3 in the supplementary material–itself energy systems [17,51]. requires particularly large amounts of water inputs per unit of energy to generate electricity in steam turbines [52]. According to the applied 3.2.3. Land occupation data [37], the use of biogas to generate heat and electricity in CHP plants The total areas of land required to maintain the heating system are requires between 16 and 3300 times as much water per unit of heat than predicted to fall dramatically under the examined scenario, as shown in all other processes being considered. As such, even though the use of Fig. 7. Indeed, the area required in 2050 is close to one third of the biogas is only predicted to rise from 0.2 % to 15.4 % of total heat energy, amount required in 2019, falling by some 64.9 %. Analysis at the three its impact on water requirements here is substantial, suggesting the levels clearly demonstrates that these reductions are almost exclusively potential risks associated with increasing biogas use. That being said, it linked to the use of wood, which contributes a mere 5.1 % of total heat must be stated that the LCI data used to define biogas here is the only by 2050, down from 37.9 % in 2019. It is also notable that the majority available listing in Ecoinvent and represents biogas from manure. It is of the land requirements in 2019 and 2050–totalling 85.5 % and 86.5 %, not known if biogas from other sources–e.g., from municipal or agri­ respectively–are linked to direct heat production from fuels, predomi­ cultural waste or sewage sludge–would return substantially different nantly from wood. Indeed, wood production is clearly the overwhelming results. factor regarding land occupation metrics in general. Meanwhile, renewable energy sources contribute less than 1 % of the total Fig. 7. Comparison of historical and predicted values for land occupation. Results are shown across three separate hierarchical levels. 8 N. Martin et al. Energy 290 (2024) 130264 Fig. 8. Comparison of historical and predicted values for water depletion. Results are shown across three separate hierarchical levels. 3.2.5. Material supply risk biogas use. Indeed, biogas provides around 49.6 % of the total score in The overall level of raw material supply risk is forecast to more than 2050 and its per-unit value for electricity is around 2.42 times higher double between 2019 and 2050, rising by 107.2 % under this scenario, than all other technologies analysed at the “n-4” level (see Table S3 in as shown in Fig. 9. Inspecting the results obtained at the “n-2” and “n-3” the supplementary material); this is strongly linked to a higher levels indicates that this growth is strongly linked to electricity gener­ requirement for platinum group metals (PGMs)–particularly rhodium ated from renewable energy technologies, although bioenergy continues and platinum–and to gold in some cases. Wind and solar power are again to be a factor. Closer inspection at the “n-4” level confirms that wind, seen to be important contributors here, occupying 32.5 % and 15.0 % of solar and biogas are the key contributors to the increase, representing the remaining share, respectively. 58.9 %, 12.9 % and 19.5 % of the total in 2050, respectively. As with most processes, neodymium, praseodymium and samarium significantly 4. Discussion influence the three key technological groups. Gadolinium and lanthanum are also notable contributors to most SR scores, gallium has a 4.1. Results and implications considerable influence on the score for solar power, while phosphorus requirements elevate the overall score for biogas. The investigation revealed a balanced spread in results in that re­ ductions were predicted in five of the indicators examined, while the 3.2.6. Environmental impacts relating to material supply remaining six indicators were predicted to increase. The complete An especially sharp rise is observed in the environmental impacts phasing out of fossil fuels and nuclear power and drastic reductions in derived from raw material extraction and processing in the case study, as wood biomass use assumed in the EnergyPLAN scenario for 2050 shown in Fig. 10. In fact, overall values are predicted to quadruple be­ resulted in beneficial reductions in the GHG emissions and fossil tween 2019 and 2050, rising by over 329 %. As with the water depletion depletion indicators, which drop by 38.0 % and 49.6 %, respectively. At indicator, this increase is connected to both direct fuel combustion and the same time, lower use levels of wood biomass are largely responsible electricity generation and is strongly linked to renewable energy and for beneficial reductions in the land occupation, marine eutrophication Fig. 9. Comparison of historical and predicted values for material supply risk. Results are shown across three separate hierarchical levels. 9 N. Martin et al. Energy 290 (2024) 130264 Fig. 10. Comparison of historical and predicted values for environmental impacts relating to material supply. Results are shown across three separate hierarchi­ cal levels. and human toxicity indicators, which fell by 64.9 %, 45.3 % and 47.5 %, 4.2. Limitations and uncertainties respectively. This again confirms the importance of reducing the use of fossil fuels as quickly as possible in order to meet Sweden’s future While the results of the analysis provide a series of useful observa­ emissions targets and the impacts of climate change both locally and tions, several minor limitations are also identified. Firstly, an issue of globally. It also provides further evidence of the benefits of curtailing uncertainty is acknowledged in relation to the ratio of electricity inputs biomass use in future scenarios. to heat outputs–the coefficient of performance (COP)–assumed in the Meanwhile, large overall reductions in water depletion amounts heat pumps considered in the study. In the data for historical systems caused by the assumed phasing out of nuclear power were over­ [6], a COP of 2.0 was assumed when calculating the heat outputs shadowed by even greater requirements coming from the use of biogas, derived from electricity use outside of district heating systems; empirical resulting in a dramatic overall increase of 251.7 % in this category. data is used for heat from electricity within district heating systems, Similarly, reductions in freshwater eutrophication potential relating to where COP values between 3.9 and 4.2 were observed. Alternatively, coal and biomass use were more than nullified by the influence of electricity requirements in EnergyPLAN are calculated using a “blanket” emerging technologies, particularly of biogas, once again, which also COP value of 4.0 for district heating; an assumed value of 3.4 is used to dominated the projected values in this category; the value for this in­ calculate heat outputs from electrical inputs in local devices, but these dicator is expected to rise by 98.7 % between 2019 and 2050. Both of values are not used in the final indicator calculations. In reality, if any of these findings suggest that the increased use of biogas production in the COP values used to calculate electricity inputs are inaccurate, the heating and other energy systems should be approached with some total electricity requirement values would change, thus affecting indi­ caution. cator calculations relating to electricity. For the remaining four indicators–supply risk, environmental im­ Furthermore, it is noted that the indicator values for producing heat pacts and environmental justice threats from raw material supply, and and electricity from biogas are calculated using the only available LCI metal depletion–any reductions relating to the phasing out of fossil fuels, listings, both of which assume that biogas derived from manure waste is wood biomass and nuclear power are predicted to be strongly offset by used to co-generate the two final energy carriers. At present, no data is the detrimental characteristics of the technologies predicted to replace available for the use of other sources of biogas and it is unclear if other them. In fact, for all four of these indicators, the dominant contributions production techniques would yield observably different results. As such, in 2050 are clearly produced by three particular technologies: solar PV, the results included here do not represent a definitive account of the wind and biogas. The use of wind turbines represents over half of the threats introduced by all biogas-related technologies; additional calcu­ overall contributions to the fossil depletion, supply risk and environ­ lations could be used to provide further insights if and when newer data mental justice indicators in 2050, resulting in large overall increases of becomes available. 110.8 %, 107.2 % and 139.3 %, respectively. Elsewhere, biogas is found Lastly, it is recognised that the calculations performed here for to be the overwhelming individual contributor to the environmental electrically generated heat do not include the life cycle impacts relating impacts from material supply indicator, contributing almost 50 % of the to the infrastructure items required to transform electricity into heat at total 2050 value to result in an overall quadrupling of the 2019 value. both the district heating and local levels. Likewise, the infrastructure Collectively, the results clearly highlight a number of situations in which required to facilitate the transfer of recycled heat into district heating the adoption of wind, solar and biogas technologies could very easily networks has not been included. Although much of this infrastructure is exacerbate environmental and material supply pressures if they are to be common to the different scenarios being considered, it is acknowledged adopted at much wider scales in future systems. They also highlight, that more specialised infrastructure is likely to be required to implement once again, the potential issues resulting from greater biogas imple­ the electrification of future systems, particularly with regards to mentation. Furthermore, if Sweden is to significantly increase the increased heat pump capacities. amount of electricity use in its heating system, and if much of this is to come from wind turbines, policymakers also need to be mindful of the potential issues that wind technologies can introduce. 10 N. Martin et al. Energy 290 (2024) 130264 5. Conclusions 873111). It also received financial support from the María de Maeztu Programme for Units of Excellence of the Spanish Ministry of Science, Buildings in Sweden require heating for at least eight months in Innovation and Universities (MICINN) (CEX2019-000940-M), the every year. The last 50 years has seen a dramatic shift away from Secretaria General de Recerca i Universitats from the Recerca i Uni­ obtaining this heat from oil heaters at the individual building level to­ versitats department of the Generalitat de Catalunya (2021SGR00734) wards centralised district heating approaches and, more recently, to­ and from the Talent UAB Santander program. wards the use of electric heat pumps. To investigate the ongoing transition in the Swedish heating system, a newly derived workflow that Appendix A. Supplementary data assesses all sources of district heating and local heat generation throughout their entire life cycles was used to identify and analyse Supplementary data to this article can be found online at https://doi. projected changes for a group of 11 key indicators. The analysis used org/10.1016/j.energy.2024.130264. system configuration data for 2015 and 2019 as historical baselines, and a predicted configuration for 2050 derived from the EnergyPLAN model. References Ultimately, the investigation provides a selection of novel and potentially concerning insights into some of the lesser-known conse­ [1] Johansson P. Heat pumps in Sweden – a historical review. Energy 2021;229: quences of heat generation practices. In particular, the benefits of 120683. https://doi.org/10.1016/j.energy.2021.120683. [2] Werner S. District heating and cooling in Sweden. Energy 2017;126:419–29. reducing the use of wood biomass–and, hence, the benefits of avoiding https://doi.org/10.1016/j.energy.2017.03.052. policies that promote wood-based technologies–could potentially be [3] Gross R, Hanna R. Path dependency in provision of domestic heating. Nat Energy offset by limitations imposed by certain wind, solar and biogas tech­ 2019;4:358–64. https://doi.org/10.1038/s41560-019-0383-5. [4] Magnusson D. Who brings the heat? - from municipal to diversified ownership in nologies. Above all, it is recognised that, although a wider range of the Swedish district heating market post-liberalization. Energy Res Social Sci 2016; modelling options is becoming available [53], policymakers must 22:198–209. https://doi.org/10.1016/j.erss.2016.10.004. continue to juggle a variety of complex issues when planning the energy [5] di Lucia L, Ericsson K. Low-carbon district heating in Sweden - examining a successful energy transition. Energy Res Social Sci 2014;4:10–20. https://doi.org/ systems of the future. And, as the pathways of the energy transition 10.1016/j.erss.2014.08.005. continue to be defined, finding compromises between the pros and cons [6] Werner S. Personal correspondence. 2022. offered by different policy options will not always be easy [54]. This will [7] Sandberg E, Sneum DM, Trømborg E. Framework conditions for Nordic district heating - similarities and differences, and why Norway sticks out. Energy 2018; certainly be the case in colder climates where implementing and 149:105–19. https://doi.org/10.1016/j.energy.2018.01.148. maintaining reliable and efficient heating systems will continue to be [8] Ericsson K, Werner S. The introduction and expansion of biomass use in Swedish vital despite the fact that GHG emissions targets and a range of other district heating systems. Biomass Bioenergy 2016;94:57–65. https://doi.org/ factors need to be satisfied. In this sense, it is hoped that deeper analyses 10.1016/j.biombioe.2016.08.011. [9] Fischer D, Madani H. On heat pumps in smart grids: a review. Renew Sustain of this kind can be used to complement existing modelling techniques Energy Rev 2017;70:342–57. https://doi.org/10.1016/j.rser.2016.11.182. and expand the scope of available tools for assisting heating policy de­ [10] Pospíšil J, Špiláček M, Kudela L. Potential of predictive control for improvement of cisions as we strive to achieve more sustainable energy systems. seasonal coefficient of performance of air source heat pump in Central European climate zone. Energy 2018;154:415–23. https://doi.org/10.1016/j. energy.2018.04.131. CRediT authorship contribution statement [11] Paardekooper S, Lund RS, Mathiesen BV, Chang M, Petersen UR, Grundahl L, et al. Heat roadmap Europe 4: Quantifying the impact of low-carbon heating and cooling roadmaps. Aalborg: Aalborg Universitetsforlag; 2018. Nick Martin: Writing - review & editing, Writing - original draft, [12] Süsser D, Martin N, Stavrakas V, Gaschnig H, Talens-Peiró L, Flamos A, et al. Why Visualization, Methodology, Formal analysis, Data curation, Conceptu­ energy models should integrate social and environmental factors: assessing user alization. Jakob Zinck Thellufsen: Writing - review & editing, Formal needs, omission impacts, and real-world accuracy in the European Union. Energy Research and Social Science Social Science 2022;92:102775. https://doi.org/ analysis, Data curation. Miguel Chang: Writing - review & editing, Data 10.1016/j.erss.2022.102775. curation. Laura Talens-Peiró: Writing - review & editing, Project [13] Chang M, Lund H, Thellufsen JZ, Østergaard PA. Perspectives on purpose-driven administration, Methodology, Funding acquisition, Data curation. coupling of energy system models. Energy 2023;265:126335. https://doi.org/ 10.1016/j.energy.2022.126335. Cristina Madrid-López: Writing - review & editing, Project adminis­ [14] Martin N, Talens-Peiró L, Villalba-Méndez G, Nebot-Medina R, Madrid-López C. An tration, Methodology, Funding acquisition, Conceptualization. energy future beyond climate neutrality: Comprehensive evaluations of transition pathways. Appl Energy 2023;331:120366. https://doi.org/10.1016/j. apenergy.2022.120366. Declaration of competing interest [15] Astudillo MF, Vaillancourt K, Pineau P, Amor B. Integrating energy system models in life cycle management. In: Benetto E, Gericke K, Guiton M, editors. Designing The authors declare that they have no known competing financial sustainable technologies, products and policies. Cham: Springer; 2018. p. 249–59. interests or personal relationships that could have appeared to influence https://doi.org/10.1007/978-3-319-66981-6_28. [16] Luderer G, Pehl M, Arvesen A, Gibon T, Bodirsky BL, de Boer HS, et al. the work reported in this paper. Environmental co-benefits and adverse side-effects of alternative power sector decarbonization strategies. Nat Commun 2019;10:5229. https://doi.org/10.1038/ Data availability s41467-019-13067-8. [17] Lamers P, Ghosh T, Upasani S, Sacchi R, Daioglou V. Linking life cycle and integrated assessment modeling to evaluate technologies in an evolving system Data will be made available on request. context: a power-to-hydrogen case study for the United States. Environ Sci Technol 2023;57:2464–73. https://doi.org/10.1021/acs.est.2c04246. [18] Weidner T, Galán-Martín Á, Walbech Ryberg M, Guillén-Gosálbez G. Energy Acknowledgements systems modeling and optimization for absolute environmental sustainability: current landscape and opportunities. Comput Chem Eng 2022;164:107883. The research received funding from the European Union’s Horizon https://doi.org/10.1016/j.compchemeng.2022.107883. [19] Laurent A, Espinosa N, Hauschild MZ. LCA of energy systems. In: Hauschild MZ, 2020 research and innovation program under the Sustainable Energy Rosenbaum RK, Olsen SI, editors. Life cycle assessment: theory and practice. Cham: Transitions Laboratory (SENTINEL) project (GA 837089), from the Springer; 2018. p. 633–68. https://doi.org/10.1007/978-3-319-56475-3_26. Spanish Ministry of Science, Innovation and Universities (MICINN) [20] Pauliuk S, Arvesen A, Stadler K, Hertwich EG. Industrial ecology in integrated assessment models. Nat Clim Change 2017;7:13–20. https://doi.org/10.1038/ under the project Living Lab on Environmental Assessment for the En­ nclimate3148. ergy Transition (LIVEN) (PID2020-119565RJ-I00) and as part of the [21] Pehl M, Arvesen A, Humpenöder F, Popp A, Hertwich EG, Luderer G. Digital Platform for Circular Economy in Cross-sectorial Sustainable Understanding future emissions from low-carbon power systems by integration of Value Networks (DigiPrime) project which is funded under the European life-cycle assessment and integrated energy modelling. Nat Energy 2017;2:939–45. https://doi.org/10.1038/s41560-017-0032-9. Union’s Horizon 2020 framework programme, DT-ICT-07-2018-2019 [22] Sacchi R, Terlouw T, Siala K, Bauer C, Cox B, Daioglou V, et al. PRospective “Digital Manufacturing Platforms for Connected Smart Factories” (ID EnvironMental Impact asSEment (premise): a streamlined approach to producing 11 N. Martin et al. Energy 290 (2024) 130264 databases for prospective life cycle assessment using integrated assessment models. [38] Goedkoop M, Heijungs R, Huijbregts M, de Schryver A, Struijs J, van Zelm R. Renew Sustain Energy Rev 2022;160:112311. https://doi.org/10.1016/j. ReCiPe 2008: a life cycle assessment method which comprises harmonised category rser.2022.112311. indicators at the midpoint and the endpoint level. Amsterdam: Ministry of Housing, [23] Baumgärtner N, Deutz S, Reinert C, Nolzen N, Kuepper LE, Hennen M, et al. Life- Spatial Planning and Environment (VROM); 2013. First edition (version 1.08). cycle assessment of sector-coupled national energy systems: environmental impacts Report I: Characterisation. of electricity, heat, and transportation in Germany till 2050. Front Energy Res [39] Martin N, Madrid-López C, Villalba-Méndez G, Talens-Peiró L. New techniques for 2021;9:621502. https://doi.org/10.3389/fenrg.2021.621502. assessing critical raw material aspects in energy and other technologies. Environ [24] Bartolozzi I, Rizzi F, Frey M. Are district heating systems and renewable energy Sci Technol 2022;56:17236–45. https://doi.org/10.1021/acs.est.2c05308. sources always an environmental win-win solution? A life cycle assessment case [40] Talens-Peiró L, Martin N, Villalba-Méndez G, Madrid-López C. Integration of raw study in Tuscany. Italy. Renewable and Sustainable Energy Reviews 2017;80: materials indicators of energy technologies into energy system models. Appl 408–20. https://doi.org/10.1016/j.rser.2017.05.231. Energy 2022;307:118150. https://doi.org/10.1016/j.apenergy.2021.118150. [25] Karlsson J, Brunzell L, Venkatesh G. Material-flow analysis, energy analysis, and [41] Swedish Energy Agency. Electricity supply, district heating and supply of natural partial environmental-LCA of a district-heating combined heat and power plant in gas 2015. Final statistics 2015. Sweden. Energy 2018;144:31–40. https://doi.org/10.1016/j.energy.2017.11.159. [42] GWEC. Global wind report 2021. Brussels: Global Wind Energy Council; 2021. [26] Guillén-Lambea S, Carvalho M, Delgado M, Lazaro A. Sustainable enhancement of [43] Connolly D, Lund H, Mathiesen BV. Smart Energy Europe: the technical and district heating and cooling configurations by combining thermal energy storage economic impact of one potential 100% renewable energy scenario for the and life cycle assessment. Clean Technol Environ Policy 2021;23:857–67. https:// European Union. Renew Sustain Energy Rev 2016;60:1634–53. https://doi.org/ doi.org/10.1007/s10098-020-01941-9. 10.1016/j.rser.2016.02.025. [27] Pakere I, Feofilovs M, Lepiksaar K, Vı̄toliņš V, Blumberga D. Multi-source district [44] Lund H, Thellufsen JZ, Østergaard PA, Sorknæs P, Skov IR, Mathiesen BV. heating system full decarbonization strategies: technical, economic, and EnergyPLAN - advanced analysis of smart energy systems. Smart Energy 2021;1: environmental assessment. Energy 2023;285:129296. https://doi.org/10.1016/j. 100007. https://doi.org/10.1016/j.segy.2021.100007. energy.2023.129296. [45] sEEnergies. Quantification of synergies between Energy Efficient First Principle [28] Jeandaux C, Videau JB, Prieur-Vernat A. Life cycle assessment of district heating and renewable energy systems for 2050 decarbonisation. 2022. https://www.seene systems in Europe: case study and recommendations. Sustainability 2021;13: rgies.eu/. [Accessed 25 April 2023]. 11256. https://doi.org/10.3390/su132011256. [46] Chesser M, Lyons P, O’Reilly P, Carroll P. Air source heat pump in-situ [29] SENTINEL. Sustainable Energy Transitions Laboratory (SENTINEL). 2023. https performance. Energy Build 2021;251:111365. https://doi.org/10.1016/j. ://sentinel.energy/. [Accessed 17 November 2023]. enbuild.2021.111365. [30] Lund H, Thellufsen JZ. EnergyPLAN - advanced energy systems analysis computer [47] Swedish Energy Agency. Electricity supply, district heating and supply of natural model. 2020. https://doi.org/10.5281/zenodo.4017214., Version 15.1. gas 2016. Final statistics 2016. [31] Østergaard PA, Lund H, Thellufsen JZ, Sorknæs P, Mathiesen B v. Review and [48] Swedish Energy Agency. Electricity supply, district heating and supply of natural validation of EnergyPLAN. Renew Sustain Energy Rev 2022;168:112724. https:// gas 2017. Final statistics 2017. doi.org/10.1016/j.rser.2022.112724. [49] Swedish Energy Agency. Electricity supply, district heating and supply of natural [32] Giampietro M, Mayumi K, Ramos-Martin J. Multi-scale integrated analysis of gas 2018. Final statistics 2018. societal and ecosystem metabolism (MuSIASEM): theoretical concepts and basic [50] Infrastrukturdepartementet. Sweden’s integrated national energy and climate plan rationale. Energy 2009;34:313–22. https://doi.org/10.1016/j. reporting. 2020. energy.2008.07.020. [51] Mendoza Beltran A, Cox B, Mutel C, van Vuuren DP, Font Vivanco D, Deetman S, [33] Swedish Energy Agency. Electricity supply, district heating and supply of natural et al. When the background matters: using scenarios from integrated assessment gas 2019. Final statistics 2019. models in prospective life cycle assessment. J Ind Ecol 2020;24:64–79. https://doi. [34] Swedish Energy Agency. Energy in Sweden facts and figures 2022. 2022. https org/10.1111/jiec.12825. ://www.energimyndigheten.se/495a34/globalassets/statistik/energilaget/energy- [52] Macknick J, Newmark R, Heath G, Hallett KC. Operational water consumption and in-sweden-facts-and-figures_220329.xlsx. [Accessed 28 August 2022]. withdrawal factors for electricity generating technologies: a review of existing [35] EnergyPLAN. EnergyPLAN model 16.22 results: sEEnergies smart energy scenario literature. Environ Res Lett 2012;7:045802. https://doi.org/10.1088/1748-9326/ for Sweden (SE_sEE_2050). 2023. 7/4/045802. [36] Wernet G, Bauer C, Steubing B, Reinhard J, Moreno-Ruiz E, Weidema B. The [53] Chang M, Thellufsen JZ, Zakeri B, Pickering B, Pfenninger S, Lund H, et al. Trends ecoinvent database version 3 (part I): overview and methodology. Int J Life Cycle in tools and approaches for modelling the energy transition. Appl Energy 2021; Assess 2016;21:1218–30. https://doi.org/10.1007/s11367-016-1087-8. 290:116731. https://doi.org/10.1016/j.apenergy.2021.116731. [37] Ecoinvent. Ecoinvent version 3.8 (2021) database. 2021. https://ecoquery.ecoin [54] Martin N, Madrid-López C, Villalba-Méndez G, Talens-Peiró L. Overlooked factors vent.org/. in predicting the transition to clean electricity. Environ Res: Infrastructure and Sustainability 2022;2:021005. https://doi.org/10.1088/2634-4505/ac70f7. 12