1 Introduction

The continuously increasing role and number of conducted life cycle assessment (LCA) studies, coupled with growing social awareness, long-term economic goals and environmental protection regulations, necessitate a deeper understanding of the methods used in life cycle impact assessment (LCIA) among practitioners. To meet this challenge, previous reviews of LCIA methods using various case studies have been conducted. However, considering recent methodological developments in LCA, further research is necessary.

The international standards LCA ISO 14040 and ISO 14044 do not specify a single particular method for performing LCIA as such. However, they provide a series of recommendations for the selection of impact categories, category indicators and characterisation models. These recommendations include international acceptance level, scientific and technological validation and environmental relevance (ISO 2006a, b, 2017 2020a, b; Finkbeiner et al. 2006). Simultaneously, the European Commission, building upon the existing standards, formulated a recommended practice for characterisation modelling in LCIA, known as ILCD 2009/2011 (EC-JRC-IES 2010, 2011, 2012). Although various LCIA methods have been developed and are available for the assessment of environmental impacts in LCA, practice shows that ReCiPe, followed by ILCD 2011, CML 2012 and IMPACT 2002+, are the most commonly used multi-impact LCIA methods at the midpoint level worldwide (Wahl 2018).

Differences in methodological approaches to characterisation modelling and the substance coverage by individual impact categories are, as previous studies have shown, influential on the final results of LCA. A comparative LCA of four types of windows for residential buildings, conducted using the IMPACT 2002+, ReCiPe 2008 and ILCD 2009 methods by the Joint Research Centre (JRC) of the European Commission, is significant in this respect (Owsianiak et al. 2014; EC-JRC-IES 2010, 2012; Hauschild et al. 2013). The research has shown though significant differences in results in the following impact categories: toxic impacts, ionising radiation, land use and mineral/metal depletion. It did not cover the LCIA optional elements, such as normalisation and weighting, which would most likely also reveal significant discrepancies in the results stemming from the selection of the LCIA method. Additionally, earlier versions of the IMPACT 2002+, ReCiPe 2008 and ILCD 2009 methods were used in the study, which now have their characterisation model updates. However, updates of LCIA methods could change the interpretation of some results of the LCA research. This is the case, for instance, with ReCiPe 2008 and its latest release – ReCiPe 2016 (Dekker et al. 2020).

Another example of research addressing methodological choices during LCIA is the comparison of the EDIP 97, CML 2001 and Eco-indicator 99 methods based on a case study of UV water-based lacquer (Dreyer et al. 2003). This study demonstrated that the choice of the LCIA method does not have a significant impact on the final results in the majority of impact categories, except for those related to toxicity. The large differences observed in the toxicity impact categories revealed variations in the characterisation modelling and factors between the CML 2001 and EDIP 97 methods. An important challenge in this comparison arises from different levels of characterisation – midpoints (EDIP 97 and CML 2001) and endpoints (Eco-indicator 99). Such selection of LCIA methods complicates the understanding of methodological differences between LCIA methods.

The comparison of seven LCIA methods, i.e. CML 2001 (midpoint), IMPACT 2002+ (midpoint and endpoint), EDIP 2003 (midpoint), Eco-indicator 99 (endpoint), TRACI 2 (midpoint), ReCiPe 2008 (midpoint and endpoint) and Ecological Scarcity 2006 (endpoint), as illustrated by a case study of gasoline and ethanol from sugarcane in Brazil, showed how the use of different LCIA methods affects the final results at both the midpoint and endpoint level (Cavalett et al. 2013). The conducted research demonstrated that ReCiPe 2008, EDIP 2003 and IMPACT 2002+ methods take a more holistic view of environmental impacts than CML 2001 and TRACI 2 methods; however, results of environmental impacts obtained using different LCIA methods at the midpoint level for equivalent impact categories are essentially similar. Consequently, using five LCIA midpoint methods, ethanol has been proven to perform better than gasoline in the following impact categories: global warming, fossil fuel depletion, ecotoxicities and ozone layer depletion. However, it does perform worse than gasoline in acidification, eutrophication, photochemical oxidation and agricultural land use. At the same time, further testing using four endpoint LCIA methods revealed large discrepancies in the results regarding areas of protection, which consequently led to different conclusions.

Despite the fact that comparative LCA studies based on different combinations of LCIA methods and various case studies have previously been conducted, showing differences in LCA results and their interpretation ranging from small to significant, depending on the choice of method and cases studied (Owsianiak et al. 2014; Borghesi et al. 2022; Silva et al. 2013; Wu and Su 2020), comprehensive studies addressing large data inventories and incorporating recent advances in LCIA methods are still very much limited. Currently, with a wide range of mid- and endpoint LCIA methods available, the choice between them is no longer obvious (Finnveden et al. 2009).

The aim of this research was to systematically compare the characterisation results based on ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+ methods, and to discuss the implications of methodological choices during LCIA for the interpretation of LCA results from the perspective of practitioners, thereby building upon previous reviews of LCIA methods. For the comparison, a case study of the electricity consumption in the 27 member states of the European Union (the EU-27), Norway, Switzerland and the United Kingdom was conducted, using different LCIA methods. Despite the fact that energy management is the top priority in the prevention of CO2 emissions, to the best of our knowledge, this is the first attempt to analyse the relationship between the inventory analysis (life cycle inventory, LCI) and LCIA results using different characterisation models for the electricity consumption mix. The choice of LCIA methods was dictated by their popularity and thus the research supports practitioners to better understand the significance, uncertainty and sensitivity of their methodological choices.

2 Methods

For the comparison of LCIA results using ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+ methods, a cradle-to-gate LCA model was developed for the electricity consumption mix in the EU-27, Norway, Switzerland and the United Kingdom in 2021. LCA is governed by ISO 14040:2006/Amd 1:2020 (Environmental management—Life cycle assessment—Principles and framework) and ISO 14044:2006/Amd 1:2017 and Amd 2:2020 (Environmental management—Life cycle assessment—Requirements and guidelines). Therefore, the research was entirely conducted in accordance with the aforementioned standards with a few necessary adjustments regarding the choice of the functional unit, the scope and the LCIA methods (Fig. 1).

Fig. 1
figure 1

Source: own study

Design of the methodological approach for the comparison of ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+ LCIA methods

Full size image

2.1 Goal and scope

LCA is a relative approach, structured around a functional unit. A unit of electricity of \(1\;MJ\) (megajoule) is used as a functional unit in the presented research. This is a derived unit within the International System of Units (SI), where \(1\;MJ\;=\;{10}^{6}\;J\) (joule), and \(J\;=\;kg\cdot {m}^{2}\cdot {s}^{-2}\). Approximately, \(1\;MJ\) corresponds to \(\sim 0.278\;kWh\) (kilowatt-hour). For tharity in data presentation, a larger prefix of the SI system, petajoule (\(PJ\)), was introduced, where \(1\, PJ={10}^{15}\;J\) (BIPM 2019).

The system boundary of the research encompasses processes from the extraction of raw materials, through the production of electricity, until the introduction of electricity to the power grid (cradle-to-gate model) in the EU-27, i.e. Austria (AT), Belgium (BE), Bulgaria (BG), Croatia (HR), Cyprus (CY), the Czech Republic (CZ), Denmark (DK), Estonia (EE), Finland (FI), France (FR), Germany (DE), Greece (GR), Hungary (HU), Ireland (IE), Italy (IT), Latvia (LV), Lithuania (LT), Luxembourg (LU), Malta (MT), the Netherlands (NL), Poland (PL), Portugal (PT), Romania (RO), Slovakia (SK), Slovenia (SI), Spain (ES) and Sweden (SE), and the EU-27 associated countries – Norway (NO), Switzerland (CH) and the United Kingdom (GB).

2.2 Life cycle inventory

LCI encompasses data on electricity consumption mix by sector, including industry, transport, commercial and public services and households, and by source, including coal, natural gas, oil, nuclear, biofuels and waste and hydro as well as other renewable sources (wind, solar, etc.) in the EU-27, Norway, Switzerland and the United Kingdom in 2021. The inventory data were sourced from the European Statistical Office (Eurostat), the Swiss Federal Office of Energy (SFOE), the UK Office for National Statistics (ONS) and the International Energy Agency (IEA). The LCI phase was augmented by datasets on electricity available in the ecoinvent database (Ecoinvent Centre 2022; Steubing et al. 2016; Frischknecht et al. 2019).

2.3 Life cycle impact assessment

2.3.1 Selection of LCIA methods and impact categories

The results of the research are presented using four LCIA methods at the midpoint level: ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+. The choice of LCIA methods was dictated by the global trends in the LCA research, revealed by a search for relevant scientific articles utilising LCIA methods, published between 2020 and 2023 (inclusive) on the Google Scholar platform (Fig. 2). The search results confirmed the findings of former studies (Wahl 2018) that ReCiPe 2016 followed by ILCD 2011, CML-IA baseline and IMPACT 2002+ are the most widely used multi-impact LCIA methods at the midpoint level worldwide, though ILCD 2011 has been recently losing on popularity.

Fig. 2
figure 2

Source: own study based on (Google Scholar 2024)

LCIA methods search results on Google Scholar

Full size image

The selected LCIA methods vary in scope, coverage and level of details (Table 1). The ReCiPe 2016 method was developed based on two methods: CML-IA, oriented towards midpoints, and Eco-Indicator 99, oriented towards endpoints. This development was a collaborative effort involving the National Institute for Public Health and the Environment (RIVM) at the Ministry of Health, Welfare and Sport in the Netherlands, Radboud University Nijmegen, Leiden University and Pré Consultants (Huijbregts et al. 2016, 2017). Therefore, it integrates both the midpoint and endpoint approach within a consistent framework encompassing 18 impact categories and three endpoint areas of protection, i.e. damage to human health, damage to ecosystem quality and damage to resource availability. The ILCD 2011 method was developed by the European Commission, Joint Research Centre (JRC), Institute for Environment and Sustainability to collect and harmonise the best recommended LCIA practices (EC-JRC-IES 2011; PRé 2020). As a result, it is grounded in the most commonly used LCIA methods and aims to introduce baseline characterisation models and factors for research purposes. The ILCD 2011 method presents the LCIA results using 16 impact categories at the midpoint level. The CML-IA method was developed by the Institute of Environmental Science (CML) of Leiden University in the Netherlands (CML-IE 2016; Guinée et al. 2002; PRé 2020). This is one of the first LCIA methods and forms the basis for the development of other LCIA methods, including ReCiPe. As part of the CML-IA method, two versions have been developed, differing in the scope of defined impact categories: CML-IA baseline, the basic version used for the research using 11 impact categories at the midpoint level to present the LCIA results, and CML-IA non-baseline, an extended version with 50 midpoint impact categories. The IMPACT 2002+ method, another of the most widely used LCIA methods, was originally developed by the Swiss Federal Institute of Technology in Lausanne (EPFL), Switzerland, and is currently maintained by the IMPACT Modelling Team (Jolliet et al. 2003; Humbert et al. 2012). Developed on the basis of two methods, CML-IA, oriented towards midpoints, and Eco-Indicator 99, a damage-oriented method, IMPACT 2002+ presents the LCIA results using 15 impact categories at the midpoint level and four damage categories, i.e. human health, ecosystem quality, climate change and resources.

Table 1 A set of impact categories for the ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+ LCIA methods (presented in alphabetical order)
Full size table

Based on the aforementioned LCIA methods, common midpoint areas of impacts were developed, grouping the midpoint impact categories upon their definitions and/or impact category units. The midpoint areas of impacts were defined for at least two convergent impact categories from among all researched methods. The resulting 12 midpoint areas of impacts are as follows: climate change, ozone depletion, ionising radiation, ozone formation/respiratory organics, particulate matter/respiratory inorganics, acidification, eutrophication, ecotoxicity, human toxicity, land use, resource scarcity (minerals and fossils) and water consumption. Only four defined midpoint areas of impacts combine coherent impact categories and units: climate change and ozone depletion (regardless of LCIA method), particulate matter/respiratory inorganics (for three methods, excluding CML-IA baseline) and water consumption (only for ReCiPe 2016 and ILCD 2011 methods). The midpoint area of impacts acidification reports results in terms of kg SO2 eq, except for the ILCD 2011 method, which reports results in terms of molc H+ eq. In the remaining 7 midpoint areas of impacts, i.e. ionising radiation, ozone formation/respiratory organics, eutrophication, ecotoxicity, human toxicity, land use and resource scarcity (minerals and fossils), there are greater discrepancies in how impact categories and units are defined.

2.3.2 Conversion of impact category results into common metrics

In order to compare the impact category results of different LCIA methods (see Tables 3–6 in the Electronic Supplementary Material), their results should be converted into common metrics. The method for converting original units into new common metrics was initially proposed by Dreyer et al. (2003) and subsequently developed and described by Owsianiak et al. (2014) (Eq. 1).

$${{\varvec{I}}{\varvec{S}}}_{{\varvec{j}}}\;=\;{IS}_{i}\;\times\;{UCF}_{i\to j}$$
(1)

where \({{\varvec{I}}{\varvec{S}}}_{{\varvec{j}}}\) is the impact result expressed in the unit of a new reference substance for the midpoint area of impact, \({{\varvec{I}}{\varvec{S}}}_{{\varvec{i}}}\) is the impact result expressed in original units for that impact category and \({{\varvec{U}}{\varvec{C}}{\varvec{F}}}_{{\varvec{i}}\to {\varvec{j}}}\) is the method – and impact category – specific unit conversion factor, defined as the reciprocal of a characterisation factor for the new reference substance (Eq. 2).

$${{\varvec{UCF}}}_{{\varvec{i}}\to {\varvec{j}}}\;=\;\frac{1}{{CF}_{i}}$$
(2)

where \({{\varvec{C}}{\varvec{F}}}_{{\varvec{i}}}\) is the characterisation factor for the new reference substance, expressed in original units \({\varvec{i}}\).

2.3.3 Unit conversion factors for midpoint areas of impacts

The criteria for the selection of new reference substances for midpoint areas of impacts were based on the following:

  1. (I)

    It must possess CF coefficients for all compared LCIA methods,

  2. (II)

    It must be a reference substance originally defined in at least one of the LCIA methods compared, and

  3. (III)

    It should be widely recognised and utilised within defined midpoint areas of impacts by LCA practitioners and endorsed by international research institutions (Table 2).

The exception pertains to the midpoint area of impact on resource scarcity in ReCiPe 2016, CML-IA baseline and IMPACT 2002+. These methods define resource scarcity using two distinct impact categories and units. To facilitate comparison, the midpoint areas of impacts for resource scarcity were subdivided into minerals (expressed in kg Fe eq) and fossils (expressed in kg oil eq). The use of iron equivalent provides a consistent and reliable basis for comparing resource scarcity (minerals). Even though it is not a unit defined by the compared methods, but it serves as a widely recognised substance reference in international scientific communities.

Table 2 Impact categories and unit conversion factors (UCFi→j) for the midpoint areas of impacts across LCIA methods – ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+
Full size table

3 Results and discussion

3.1 Inventory analysis

The results of the inventory analysis present electricity consumption mix in the EU-27, Norway, Switzerland and the United Kingdom by sector and source in 2021. The total electricity consumption mix equals 10,959.215PJ, with the largest share held by the following: Germany (16.94%), France (14.54%), Italy (9.87%), the United Kingdom (9.39%) and Spain (7.72%). The lowest shares were held by the following: Malta (0.08%), Cyprus (0.15%), Luxembourg (0.21%), Latvia (0.23%) and Estonia (0.28%) (see Fig. 1 in the Electronic Supplementary Material) (Eurostat 2022; SFOE 2022; BEIS 2022; ISO 2020c). The industry sector accounted for 36.20% of electricity consumption mix, the transport sector for 2.23%, the commercial and public services for 30.69% and households for 30.88%. Germany, France, Italy, the United Kingdom and Spain had the greatest share in electricity consumption mix, irrespective of the analysed sector. The only exception was the transport sector, where Germany, Italy, France, the United Kingdom and Switzerland had the greatest share in electricity consumption mix among the analysed countries. The lowest electricity consumption mix in the industry sector and households is recorded in Malta, Cyprus, Luxembourg, Latvia and Estonia, in the transport sector Malta, Latvia, Lithuania, Estonia and Ireland and in commercial and public services Malta, Latvia, Cyprus, Luxembourg and Slovenia (Eurostat 2022; SFOE 2022; BEIS 2022; Smetana et al. 2015).

The electricity generation mix of the EU-27, Norway, Switzerland and the United Kingdom is very diverse, comprising coal (13.38%), natural gas (19.64%), oil (1.41%), nuclear power (23.15%), biofuels and waste (6.95%), hydro power (16.47%) and other renewable sources (wind, solar, etc.) (19.00%) (Fig. 3) (IEA 2022; Eurostat 2022; SFOE 2022; BEIS 2022; Basosi et al. 2020). The highest share in the electricity generation mix is attributed to nuclear power, which constitutes the largest portion in France (68.32%), followed by Slovakia (52.41%), Belgium (50.10%), Hungary (44.27%) and the Czech Republic (36.12%). Nuclear power has no share in the electricity generation mix of 15 countries: Austria, Croatia, Cyprus, Denmark, Estonia, Greece, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Norway, Poland and Portugal. When it comes to the electricity generation mix from renewable sources, Norway leads in hydro power (91.38% of the electricity generation mix), while Denmark, Lithuania and Spain lead in other renewable sources, including wind and solar (52.54%, 34.35% and 32.60%, respectively). Finally, Denmark, along with Estonia, Luxembourg and Finland, leads in biofuels and waste (28.78%, 25.83%, 20.73% and 19.54%, respectively).

Fig. 3
figure 3

Source: own study based on (IEA 2022)

Electricity generation mix by source in the EU-27, Norway, Switzerland and the United Kingdom in 2021

Full size image

3.2 Impact assessment

The research offered valuable insights into environmental profiles of electricity consumption mix for the EU-27, Norway, Switzerland and the United Kingdom using the most frequently used LCIA methods, i.e. ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+. The results presented below include the ranking of environmental profiles of electricity consumption mix in the analysed countries across the defined midpoint areas of impacts and a comparison of the results in absolute values across the defined midpoint areas of impacts (Fig. 4), and an explanation for differences in the obtained results based on the contribution analysis (Fig. 5).

Fig. 4
figure 4figure 4figure 4

Source: own study

Comparison of the results in absolute values and the rankings of environmental profiles of electricity consumption mix in the EU-27, Norway, Switzerland and the United Kingdom across the defined midpoint areas of impacts listed in Table 2 (see Tables 3–6 in the Electronic Supplementary Material)

Full size image
Fig. 5
figure 5figure 5

Source: own study

Results of the contribution analysis using impact scores converted into common metrics for ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+. Only substances contributing above 1% of the total impact are shown, while substances contributing <1% of total impact were merged into the substance category "Others".

Full size image

3.2.1 Climate change

Regardless of the LCIA method adopted, both the rankings of environmental profiles of the electricity consumption in the EU-27, Norway, Switzerland and the United Kingdom and the results in absolute values in the defined midpoint area of impact – climate change – are strikingly similar. This confirms high level of correlation in the modelling of this impact category between the analysed methods, which all are based on characterisation factors developed by the UN’s Intergovernmental Panel on Climate Change (IPCC). The ranking of electricity consumption in relation to climate change has generally the following order: Poland, Austria, the Czech Republic, Cyprus and Estonia on the one side and Norway, Sweden, Croatia, France and Switzerland on the other side. Consequently, Poland (0.268 – 0.280 kg CO2 eq), Austria (0.268 – 0.275 kg CO2 eq), Cyprus (0.261 – 0.265 kg CO2 eq), the Czech Republic (0.252 – 0.258 kg CO2 eq) and Estonia (0.240 – 0.247 kg CO2 eq) exhibit the highest results in climate change impact, whereas Norway (0.006 kg CO2 eq), Sweden (0.014 – 0.015 kg CO2 eq), Croatia (0.016 – 0.017 kg CO2 eq), France (0.023 – 0.024 kg CO2 eq) and Switzerland (0.027 – 0.031 kg CO2 eq) demonstrate the lowest results. Carbon dioxide, fossil (CO2), followed by methane, fossil (CH4) and dinitrogen monoxide (N2O) contribute most significantly to the results of climate change (on average, respectively accounting for approximately 95.98%, 2.07% and 0.96%). In conclusion, the choice of the LCIA method (ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+) in terms of climate change does not have a significant impact on the final LCIA results.

3.2.2 Ozone depletion

Both the rankings of environmental profiles of electricity consumption in the EU-27, Norway, Switzerland and the United Kingdom and the results in absolute values in the defined midpoint area of impact – ozone depletion – differ considerably between ReCiPe 2016 and the remaining LCIA methods, i.e. ILCD 2011, CML-IA baseline and IMPACT 2002+. According to ReCiPe 2016, the ranking of electricity consumption in terms of ozone depletion has the following order: Austria, Cyprus, Estonia, Malta and Latvia on the one side and Norway, Croatia, Sweden France and Romania on the other side. However, according to ILCD 2011, CML-IA baseline and IMPACT 2002+ Cyprus, Estonia, Malta, France and Belgium are on the one side and Norway, Croatia, Austria, Romania and Poland are on the other. Consequently, the most significant difference pertains to the results in ozone depletion for electricity consumption in Austria, which ranks 1st in ReCiPe 2016 and 28th in ILCD 2011, CML-IA baseline and IMPACT 2002+. On average, the results obtained using ILCD 2011, CML-IA baseline and IMPACT 2002+ are lower than those obtained using the ReCiPe 2016 method by 82.23%. And thus, according to the former three methods, the highest results in absolute values are obtained for Cyprus (4.21E−08 kg CFC-11 eq), Estonia (4.00E−08 kg CFC-11 eq), Malta (2.90E−08 kg CFC-11 eq), France (2.56E−08 kg CFC-11 eq) and Belgium (2.35E−08 kg CFC-11 eq), while the lowest for Norway (6.79E−10 kg CFC-11 eq), Croatia (1.93E−09 kg CFC-11 eq), Austria (3.20E−09–3.22E-09 kg CFC-11 eq), Romania (3.97E−09 kg CFC-11 eq) and Poland (4.31E−09 kg CFC-11 eq). Dinitrogen monoxide (84.46%) contributes most to the results of ozone depletion in the ReCiPe 2016 method, whereas methane, i.e. Halon 1301, Halon 1211 and CFC-12 (59.42%), and ethane, i.e. CFC-114 (39.06%), contribute the most to the results of ozone depletion according to the ILCD 2011, CML-IA baseline and IMPACT 2002+ methods. This is one of the reasons for the differences in results. Although all characterisation methods are based on the general consensus of the World Meteorological Organisation (WMO), ReCiPe 2016 uses newer approaches and models, specifically WMO (2011), while the other three methods (i.e. ILCD 2011, CML-IA baseline and IMPACT 2002+) are based on the WMO (1999) model.

3.2.3 Ionising radiation

Despite the fact that the results in absolute values in the defined midpoint area of impact – ionising radiation – differ considerably between ReCiPe 2016, ILCD 2011 and IMPACT 2002+, the rankings of environmental profiles of electricity consumption in the EU-27, Norway, Switzerland and the United Kingdom are nearly identical. According to ReCiPe 2016, France (1.26E−01 kBq 235U eq), Belgium (8.39E − 02 kBq 235U eq), Hungary (7.91E − 02 kBq 235U eq), Slovakia (7.56E−02 kBq 235U eq) and Sweden (6.60E−02 kBq 235U eq) have the highest results in ionising radiation, whereas Croatia (2.89E−04 kBq 235U eq), Austria (5.33E−04 kBq 235U eq), Cyprus (1.86E−03 kBq 235U eq), Norway (1.99E−03 kBq 235U eq) and Ireland (2.89 E−03 kBq 235U eq) have the lowest results. The rankings of electricity consumption in terms of ionising radiation according to the remaining two LCIA methods, i.e. ILCD 2011 and IMPACT 2002+, are similar. Therefore, France, Sweden, Belgium, Finland and Hungary are at the forefront, and in contrast, Croatia, Austria, Norway, Portugal and Poland are at the bottom of the ranking. The results in absolute values for ionising radiation using ReCiPe 2016 and ILCD 2011 are comparable and, on average, 93.39% and 90.21% lower than for IMPACT 2002+. Radon-222 (222Rn) and carbon-14 (14C) contribute the most to the results of ionising radiation, on average 99.38% in ReCiPe 2016, 99.66% in ILCD 2011 and IMPACT 2002+ 97.76%. In conclusion, the results for ionising radiation confirm that this impact category is directly linked to the use of nuclear power in the electricity mix.

3.2.4 Ozone formation/respiratory organics

Although the results in absolute values in the defined midpoint area of impact—ozone formation/respiratory organics—differ considerably between ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+ , the rankings of environmental profiles of electricity consumption in the EU-27, Norway, Switzerland and the United Kingdom are very close, with the exception of few countries. Consequently, for both ReCiPe 2016 and ILCD 2011, Cyprus, Estonia, Poland, Malta and Austria have the highest impacts in ozone formation/respiratory organics, whereas Norway, Spain, France, Hungary and Switzerland have the lowest. According to the CML-IA baseline, Cyprus, Slovenia, Estonia, Poland and Greece are at the forefront, whereas electricity consumption countries similar to those in ReCiPe 2016 and ILCD 201, i.e. Norway, Spain, France, Hungary and Switzerland, have the lowest impacts in ozone formation/respiratory organics. Finally, according to IMPACT 2002+ , Estonia, Cyprus, Latvia, Malta and Lithuania have the highest impacts, whereas Norway, Hungary, Spain, Switzerland and France have the lowest impacts in ozone formation/respiratory organics. Regarding the differences in the rankings of environmental profiles, one of the greatest considers the impacts of electricity consumption in Slovenia, ranking 11th, 8th, 2nd and 25th place following ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+ respectively. ReCiPe 2016, due to its comprehensiveness, yields highest results in absolute values for ozone formation/respiratory organics. Compared to ReCiPe 2016, the remaining methods, including ILCD 2011, CML-IA baseline and IMPACT 2002+ , give on average 89.03%, 98.49% and 99.09% lower results, respectively.

Nitrogen oxides (NOx), followed by non-methane volatile organic compounds (NMVOC) contribute the most to the categories of ozone formation/respiratory organics (on average 99.73%) in ReCiPe 2016. A similar situation occurs in the case of ILCD 2011, where NOx, NMVOC and sulphur dioxide (SO2) have the largest share (98.98%) in ozone formation/respiratory organics. In the case of the remaining two methods, ozone formation/respiratory organics largely depend on the amount of SO2 (83.93%) in CML-IA baseline or NMVOC (81.55%) in IMPACT 2002+ .

3.2.5 Particulate matter/respiratory inorganics

The rankings of environmental profiles of electricity consumption in the EU-27, Norway, Switzerland and the United Kingdom in the defined midpoint area of impact – particulate matter/respiratory inorganics – are very similar, despite the differences in the results in absolute values between ReCiPe 2016 and the remaining two LCIA methods, ILCD 2011 and IMPACT 2002+. Consequently, notwithstanding the method, electricity consumption in Cyprus, Slovenia, Estonia, Poland and Greece (or Romania instead of Slovenia in IMPACT 2002+) have the highest impacts in particulate matter/respiratory inorganics, whereas Norway, Sweden, France, Croatia and Switzerland have the lowest negative impacts. As the results in absolute values vary significantly depending on the method, ReCiPe 2016 yields the highest results for particulate matter/respiratory inorganic. The remaining methods, including ILCD 2011 and IMPACT 2002+, give on average 70.46% and 45.9% lower results compared to ReCiPe 2016. The differences in the results are related to the scope of the characterisation models – the ReCiPe 2016 method focuses on sulphur dioxide, which together with primary and secondary PM2.5 particles and nitrogen oxides contributes on average 98.68% to particulate matter/respiratory inorganics.

3.2.6 Acidification

Both the rankings of environmental profiles of electricity consumption in the EU-27, Norway, Switzerland and the United Kingdom and the results in absolute values in the defined midpoint area of impact – acidification – are almost similar for ReCiPe 2016, ILCD 2011, CML-IA baseline, whereas there are some differences for IMPACT 2002+. And thus, regardless of the LCIA method, the highest negative impacts in acidification are observed in electricity consumption in Cyprus, Slovenia, Estonia, Poland and Malta, although the order varies depending on the method. Norway, Sweden, France, Hungary and Switzerland exhibit the lowest impacts. Regarding the comparison of the results in absolute values, IMPACT 2002+ yields the highest results in acidification, which stems from the fact that it covers the most comprehensive scope of acidification, i.e. terrestrial, nutri and aquatic. Due to narrower scope, the remaining methods, including ReCiPe 2016, ILCD 2011 and CML-IA baseline, give on average 78.62%, 76.54% and 73.82% lower results compared to IMPACT 2002+ . For example, ReCiPe 2016 focuses solely on terrestrial acidification. Sulphur dioxide together with nitrogen oxides and ammonia (NH3) contributes the most to the results of acidification in all analysed methods, on average 99.93% in ReCiPe 2016, 99.95% in ILCD 2011, 99.22% in CML-IA baseline and 98.93% in IMPACT 2002+.

3.2.7 Eutrophication

The rankings of environmental profiles of electricity consumption in the EU-27, Norway, Switzerland and the United Kingdom in the defined midpoint area of impact – eutrophication – are very similar, although small differences in rankings remain, whereas the results expressed in absolute values using ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+ reveal larger discrepancies, arising from the differences in the modelling of eutrophication. In general, electricity consumption in Greece, Austria, Czech Republic, Bulgaria and Poland has the highest impacts in eutrophication, whereas that in Norway, Hungary, Cyprus, Sweden and France according to ReCiPe 2016 and in Norway, Hungary, Sweden, France and Belgium according to ILCD 2011, IMPACT 2002+ and CML-IA baseline has the lowest. Regarding the comparison of the results in absolute values, ILCD 2011 and CML-IA baseline yield the highest results in eutrophication, which stems from the fact that they cover the most comprehensive scope of eutrophication, i.e. freshwater, marine and terrestrial. Due to narrower scope, the remaining methods, including ReCiPe 2016 and IMPACT 2002+, give on average 11.97% and 90.01% lower results compared to ILCD 2011. In general, phosphate (PO43–), nitrogen oxides and nitrate (NO3–) contribute the most to the results of eutrophication in all analysed LCIA methods; however, IMPACT 2002+ focuses predominantly on emissions of phosphate to water, on average 99.96% in ReCiPe 2016, 99.86% in ILCD 2011, 98.82% in CML-IA baseline and 95.24% in IMPACT 2002+.

3.2.8 Ecotoxicity

Both the rankings of environmental profiles and the absolute values of electricity consumption mix in the EU-27, Norway, Switzerland and the United Kingdom in the defined midpoint area of impact – ecotoxicity – differ significantly among ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+. The results do not demonstrate consistency in the rankings of environmental profiles across the applied methods. The most significant discrepancies are observed in the case of countries such as Bulgaria, Greece, Cyprus, Latvia and Romania, where differences in ranking positions between all methods are 24, 24, 23, 22 and 22 places, respectively. However, minor variations in rankings are noted for Norway and Hungary, where these differences are 1 and 2 places, respectively. In the analysis of absolute values using the ReCiPe 2016, ILCD 2011 and IMPACT 2002+ methods, the results were lower by 99.86%, 99.99% and 97.25%, respectively, compared to the CML-IA baseline method. The most significant differences were observed in countries such as Poland, Austria, the Czech Republic, Greece and Bulgaria, where the results differed by almost 100% compared to the CML-IA baseline method. The key distinction lies in marine aquatic ecotoxicity for the CML-IA baseline method, where the results are significantly higher than in all other impact categories. In the contribution analysis of ReCiPe 2016, copper, vanadium, nickel, zinc, silver and mercury have the greatest impact (on average 93.32%). ILCD 2011 is dominated by copper, zinc, nickel, vanadium and chromium VI (on average 95.69%). The CML-IA baseline highlights beryllium, hydrogen fluoride, nickel, cobalt, selenium, vanadium, copper, barium and thallium (on average 98.84%). The IMPACT 2002+ emphasises aluminium, copper, zinc, chromium VI and nickel (on average 95.77%).

3.2.9 Human toxicity

Rankings of environmental profiles and absolute values for electricity consumption mix in the defined midpoint area of impact – human toxicity – in the EU-27, Norway, Switzerland and the United Kingdom exhibit only marginal similarities among the various LCIA methods ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2011. Comparisons of rankings between ReCiPe 2016, ILCD 2011 and CML-IA baseline methods revealed slight, albeit irregular, similarities, with average differences of up to 3 places. For the IMPACT 2002+ method, no similarities with the rankings of other methods were noted. Particularly significant differences occur in countries such as Latvia, Cyprus, Estonia, Bulgaria and Romania, where differences in ranking positions across all methods are 20, 19, 19, 18 and 18 places, respectively. It is noteworthy that ranking positions for Norway are identical across all four LCIA methods, with slight changes observed for Croatia and Poland. Absolute values obtained using the ReCiPe 2016, CML-IA baseline and IMPACT 2002+ methods are lower by 74.88%, 88.23% and 5.68%, respectively, compared to the ILCD 2011 method. The analysis of absolute values for all methods revealed the greatest differences in countries such as Estonia, Latvia, Lithuania, the United Kingdom and Italy, where results differed by 94.83%, 94.42%, 93.41%, 93.29% and 93.10%, respectively, between LCIA methods. In the contribution analysis of ReCiPe 2016 and ILCD 2011, zinc, arsenic, chromium VI, mercury and barium have the greatest impact (on average 94.42% and 95.15%, respectively). The CML-IA baseline highlights selenium, thallium, nickel, chromium VI, barium, hydrogen fluoride, molybdenum, vanadium and beryllium (on average 90.95%). The IMPACT 2002+ emphasises hydrocarbons – aromatic, arsenic, dioxin, benzo(a)pyrene, polycyclic aromatic hydrocarbons, zinc and barium (on average 96.17%).

3.2.10 Land use

Rankings of environmental profiles and absolute values for electricity consumption mix in the defined midpoint area of impact – land use – in the EU-27, Norway, Switzerland and the United Kingdom exhibit notable consistencies when using the ReCiPe 2016 and IMPACT 2002+ methods. However, significant disparities emerge when using the ILCD 2011 method. The analysis of ReCiPe 2016, ILCD 2011 and IMPACT 2002+ rankings reveal average variations exceeding 6 places, particularly for Cyprus, Luxembourg, Austria, Slovenia and the United Kingdom. When excluding the ILCD 2011 method and comparing only ReCiPe 2016 and IMPACT 2002+, the average differences in rankings amount to 1.5 places. Significant changes are observed for Slovenia, the United Kingdom, Finland, Spain and Sweden, with variations of 5, 5, 5, 4 and 4 places, respectively. However, France consistently occupies the 27th place in all the analysed rankings. The analysis of absolute values reveals that, on average, ReCiPe 2016 yields 37.33% higher results compared to the ILCD 2011 method and 8.29% higher results compared to the IMPACT 2002+ method. The largest divergences among analysed countries are for Norway, Cyprus, Luxembourg, Germany and Slovenia, differing by 94.30%, 87.79%, 84.39, 77.25% and 75.46%, respectively. The occupation of urban and green areas, followed by forest occupation and intensive land use, significantly influences ReCiPe 2016 and IMPACT 2002+ methods. In contrast, ILCD 2011 is dominated by factors like transformation into mineral extraction areas, industrial areas and annual crop areas. Results differ because ReCiPe 2016 defines its land use characterisation model with competitive land types, while ILCD 2011 considers both land occupation and its transformation. The contribution analysis shows that in the ReCiPe 2016 and IMPACT 2002+ methods, the largest impacts on land use are from 'occupation, urban, green areas' (77.66% and 88.11%, respectively) and 'occupation, forest, intensive' (18.25% and 6.83%, respectively). In the ILCD 2011 method, key impacts are noted in 'transformation, to mineral extraction site' (63.06%), 'occupation, forest, intensive' (10.06%), 'transformation, to industrial area' (8.73%) and 'transformation, to annual crop, non-irrigated, intensive' (3.58%).

3.2.11 Resource scarcity (minerals and fossils)

The ranking of environmental profiles and absolute values for the electricity consumption mix in the EU-27, Norway, Switzerland and the United Kingdom in the defined midpoint area of impact – resource scarcity (minerals and fossils) – varies significantly across the ReCiPe 2016, CML-IA baseline and IMPACT 2002+ methods. In the defined midpoint area of impact resource scarcity (minerals), an average difference of 7.67 places between ReCiPe 2016 and IMPACT 2002+ is observed. Ireland, Cyprus, Slovakia and Hungary experienced substantial ranking changes of 19, 18, 18 and 16 places, respectively. The absolute value analysis indicates that ReCiPe 2016 results are, on average, 66.27% lower than IMPACT 2002+. The largest disparities are observed for Norway, Ireland, Croatia, Austria and Cyprus, differing by 79.36%, 77.89%, 77.63%, 76.20% and 76.20%, respectively. In the contribution analysis of ReCiPe 2016, iron, gold, copper, uranium and nickel have the greatest impact (90.98%). The IMPACT 2002+ emphasises copper, nickel, iron and aluminium (95.82%).

Comparatively, for the defined midpoint area of impact resource scarcity (fossils), differences of 5.6 places exist between ReCiPe 2016, CML-IA baseline and IMPACT 2002+, with only a 0.2 places difference between ReCiPe 2016 and CML-IA baseline. Notably, France, Hungary, Belgium, Ireland and Latvia experience considerable ranking changes of 22, 14, 13, 11 and 11 places, respectively. Absolute value analysis reveals that ReCiPe 2016 and CML-IA baseline results are, on average, 37.88% and 38.43% lower, respectively, compared to IMPACT 2002+. The greatest disparities are for the United Kingdom, Cyprus, Denmark, Norway and Latvia, differing by 91.56%, 91.12%, 76.77%, 71.89% and 71.31%, respectively. In the contribution analysis, the largest shares were demonstrated for brown coal, crude oil, hard coal and natural gas, but with varying proportions for the ReCiPe 2016 (99.00%), CML-IA baseline (99.73%) and IMPACT 2002+ (71.85%) methods, which also consider uranium (27.95%).

The consistency in results is attributed to employing the characterisation model for the ReCiPe 2016 method, based on the CML-IA baseline model. Differences between ReCiPe 2016 and CML-IA baseline, compared to IMPACT 2002+ primarily arise from variations in the classification of uranium. Uranium is classified as minerals resource scarcity in ReCiPe 2016 and CML-IA baseline, but as fossils resource scarcity in IMPACT 2002+, significantly impacting the results.

3.2.12 Water consumption

The rankings of environmental profiles and the absolute values of the electricity consumption mix in the EU-27, Norway, Switzerland and the United Kingdom in the defined midpoint area of impact – water consumption – significantly differ between the ReCiPe 2016 and ILCD 2011 methods. When examining rankings of both methods, average differences of 5.8 places were observed between them. The most substantial disparities in rankings were identified in Norway, with a difference of 27 places, followed by Finland, Spain, Slovenia and Sweden, with differences of 15, 14, 12 and 11 places, respectively. An analysis of absolute values revealed that the ReCiPe 2016 method yields, on average, 52.63% higher results compared to the ILCD 2011 method. The most significant variations in this aspect were observed in Norway, Finland, Sweden, Slovenia and Estonia, where differences amounted to 99.47%, 93.05%, 90.79%, 88.04% and 85.97%, respectively. The ReCiPe 2016 method defines water consumption impact as the use of water through processes like evaporation, incorporation into products, transfer to other watersheds or discharge into the sea, making water unavailable in the watershed to humans and ecosystems (Huijbregts et al. 2017). In contrast, the ILCD 2011 method assesses water resource depletion by establishing a reference flow based on the weighted average consumption in the EU-27. Ecological coefficients of other water flows are then linked to express their impacts in equivalent water consumption (EC-JRC-IES 2012; Borghesi et al. 2022).

4 Conclusions

LCA is a technique of growing interest, applied to quantify the use of resources and the release of environmental impacts throughout a product’s life cycle. The LCA methodology encompasses four phases, one of which is LCIA. The phase of LCIA, in concept and methodological layout, is a process associating inventory data with specific environmental impact categories and category indicators, thereby attempting to understand these impacts (ISO 2006b, 2017, 2020b). However, several issues, including the choice of the LCIA method, as our research proved in practice, have important implications for the LCIA results and thus causing some uncertainty for the conclusions and recommendations formulated.

In the paper, we analysed four LCIA methods (ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002+) using the electricity consumption mixes in the EU-27, Norway, Switzerland and the United Kingdom as the case study. Despite the fact that these are the most frequently used multi-impact LCIA methods at the midpoint level, only two of them – ReCiPe 2016 and ILCD 2011 – are regularly updated and provide the guidance documents for practitioners.

In general, the use of the ReCiPe 2016, ILCD 2011, CML 2012 and IMPACT 2002+ methods lead to different characterisation results in absolute values, except for the following midpoint areas of impacts – climate change, ozone depletion (excluding the ReCiPe 2016 method), acidification (excluding the IMPACT 2002+ method) and eutrophication (excluding the IMPACT 2002+ method), where the choice of the LCIA method did not have significant implications for the results. Such results were expected, because of large methodological discrepancies between LCIA methods, concerning, among others, the geographical differentiation (either the European or global context), scope of the impact categories grouped into the same midpoint area of impacts (for instance eutrophication) and the contribution of specific environmental flows to a given result of the impact category. On the other hand, however, the relative results, the ranking of environmental profiles of electricity consumption mix in the analysed countries did not change to any large extent, except for ozone formation/respiratory organics, ecotoxicity, resource scarcity and water consumption. Consequently, although the comparative LCIA of the electricity consumption mix of the EU-27, Norway, Switzerland and the United Kingdom using ReCiPe 2016, ILCD 2011, CML-IA baseline and IMPACT 2002 demonstrated some trends regarding the magnitude of impacts on the environment, the decision about the best and the worst environmental profile of the individual electricity consumption mix depends on which environmental impacts are prioritised.

The findings of this research underpin the importance of careful selection of the LCIA method and in the case of observed differences in results, choosing a method that best reflects the specificity of an analysed product system, geographical context and completeness of the characterisation model, following the goal and scope definition, and interpretation of LCA results. Future research, based on the findings of this study, could unfold in several directions. One of which, the most probable, is to perform a comparative analysis of LCIA methods using different case studies, which is very likely to reveal further discrepancies between methods.