A Review on Consequential Life Cycle Assessment in the Power Sector

According to UNFCC, the greenhouse gas (GHG) emissions of energy sector accounted for 90% of the total GHG emissions of Annex I countries (including land use, land use change and forestation) in 2018, at 13.47 GtCO₂e out of 14.91 GtCO₂e [1]. The power and heat generation sector, in particular, contributed to 9.76 GtCO₂e in 2017 and increased 2.5% in 2018 [2]. The deployment of renewable energy and energy efficiency technologies help to reduce the emission intensity by 1.3%. However, the total GHG emissions of the sector generally increase due to the growing demand of power [2].

Decarbonisation, sustainable environment, economic prosperity, and social equity are the cores of the current GHG reduction policy [3]. It requires the deep decarbonisation of the economies, in which the energy system as well as other emission intensive sectors need to transform into zero emission ones, while ensuring the economic development and meeting the growing needs of population [4]. To reach the goal of decarbonisation, the industries need to reduce their GHG emission while maintaining the developing trend of the economy. The Deep Decarbonisation Pathway Project countries, which contribute to 74% of global energy-related GHG emissions, sets the objective that by 2050 the GHG emissions of energy sector will be reduced of 46-56% compared to 2010 level, while maintaining the average gross domestic product (GDP) growth of 3.1% and the population growth of 17% annually [4].

The decarbonisation process of the power sector may have some negative environmental impacts on other economic sectors. For instance, the use of bio-power may reduce the GHG emission contribution of the power sector thanks to the substitution effect for fossil-fuel-based power, but such substitution causes eutrophication and acidification due to the extensive energy crop plantation [5]. In such case, the GHG emissions of the power sector will reduce, but other negative environmental impacts within the agriculture sector will increase.

At the same time, as power and energy participate into the supply chain of every product and service, any environmental impacts of power and energy consumption or environmental benefits of energy recovery during the product life cycle will contribute to the environmental impacts or benefits of other product system. Examples in agriculture products, e.g., apples in the North of Italy, indicates that nearly 40% of climate change impacts of these products come from energy consumption during the transportation and another 40% come from power consumption during the storage of post harvesting activities [6]. An example of environmental benefits can be clearly observed in the case of waste for energy in Palermo, in which the global warming impact of conventional waste management system, with small amount of waste being classified and recycled, triples that of the waste management approaches with material and energy recovery [7].

Life cycle assessment (LCA) quantifies the life cycle environmental impacts of a product system, covering all stages, including raw material extraction and processing; product/ service manufacturing, use and disposal; and transportation [8]. LCA estimates the life cycle environmental burdens of a product system [8]. The comprehensiveness makes LCA a particularly effective mechanism for quantifying different environmental impacts originating from the product’s life cycle including indirect impacts.

There are two types of LCA approaches, namely attributional LCA (ALCA) and consequential LCA (CLCA). In the ALCA approach, inputs and outputs of a product system are attributed to its functional unit by linking the unit processes of the system while defining a physical boundary and isolating it from other systems [9]. Meanwhile, CLCA has been developed to quantify the environmental impacts of a product system in relation with changes within its life cycle [9].

Due to the principal differences between the two approaches, it is expected that the obtained results with the application of each approach will be different. This paper reviews CLCA studies in power sector in terms of the obtained results and methodology in order to consider the capability of CLCA approach to model the indirect environmental impacts of the power system and power generation/storage technologies in the context of the existing GHG policy.

3.1 System boundary

CLCA approach expands the studied system boundary by including different unit processes [18] and products and co-products [21] to the extent of the expected changes. All of the reviewed case studies expanded the product system boundary by either increasing unit processes or including relevant products and co-products. Some studies made it very clear on the inclusion of extended unit processes [22-26], while others did not clearly identify the unit processes of a product system’s life cycle [27-34]. In terms of products and co-products, the system boundary was extended to at least two products in all reviewed studies. Moreover, it was even extended to several relevant economic sectors. The investigated products and economic sectors of some reviewed CLCA papers are specified in Table 2.

There are two approaches for dealing with the expansion of system boundary in a CLCA, namely ceteris paribus (other things being equal) and mutatis mutandis (the necessary changes being made) assumptions. In the ceteris paribus assumption, the changes in the product system are analysed in isolation, as the scale of effect is small [35]. The products/ technologies, which are affected by the changes, are called marginal products/ technologies [36]. In contrast, the mutatis mutandis approach assumes that the changes in the product system are large enough to induce changes in other socio-economic sectors [37].

The first assumption was proposed by Weidema [35]. The author suggested an approach to identify marginal products/ technologies in five steps [35, 38], including:

(1) The scale and time horizon of the study,

(2) The scale of the effect, at process or market scale,

(3) The trend in volume of the affected market,

(4) The potential of an increase or reduction in supply and demand, and

(5) The favourability of the technology.

The approach clarified not only the link between the product systems and unit processes through intermediate products, which is conveyed in an ALCA, but also the consequences on supply and demand of products and co-products [38]. A study of Mattsson et al. [38] indicated that a change in the annual power demand by 1 TWh induced the same consequence on the power technologies as a change of 1 kWh. Therefore, the change in the annual power demand by 1 TWh may be regarded as small. It was suggested that in case of limitation of data availability, the scale of change was assumed to be small [38].

Several examples of ceteris paribus assumption were presented in the studies of Eriksson et al. [22] and Jones et al. [39], in which the marginal products or technologies are replaced by an alternative product (Refer to Table 2). This assumption was either applied at the early time of CLCA development, when the methodology was emerging, for example in study of Eriksson et al. [22], or in CLCA studies which focused on dealing with the physical changes of the environmental impacts, for example in Jones et al.’s study [39].

In Eriksson et al.’s study, the environmental impacts of heat and power from waste incineration were calculated and compared with other competing fuels. The assumed marginal heat and power included combined heat and power (CHP) from biomass combustion, heat from biomass combustion and CHP from natural gas combustion. These marginal products were substituted with the ratio 1:1 by heat and power from waste incineration and material recycling [22].

The study of Jones et al.’s assessed the environmental consequences of increased PV uptake in UK within 30 years by three scenarios. Corresponding to these scenarios, there were changes in energy generation technologies, distribution network structure and grid operation. These changes in energy flows were quantified by using net energy analysis (NEA), and the changes in relevant environmental consequences were quantified by using CLCA [39]. The framework of NEA and CLCA was combined by setting the targets on distributed energy uptake, specifically 50 GW PV with domestic storage, 50GW PV without domestic storage and 10 GW PV without domestic storage. These targets were set by consulting with stakeholders, including distributed generation technology suppliers and installers, electricity network and grid operators and policy makers through a workshop. Environmental impacts due to the increase of distributed power such as impacts on raw material mining, manufacturing and transport infrastructure originated from increased deployment of solar PV were cutted-off. Besides, change in power consumption patterns due to integration of distributed power into buildings were excluded in the system boundary. The exclusion of these secondary impacts reduced the complexity and uncertainty of the studies without affecting the research focus of changes in power system as a whole [39].

The reviewed papers that applied the ceteris paribus assumption focused on the causal relations of change in the product system and affected product rather than socio-economic relations between them. Therefore, these “claimed to be CLCA” studies should be regarded as using consequential concept, not a CLCA.

Frischknecht and Stucki proposed an approach to select the suitable assumption based on quantitative justifications [36]. In their approach, different modelling techniques would be applied depending on the changing agent, its potential effect and the size of studied product systems. It was suggested that if the potential consequences of the effect were small (the product system’s economic size accounts for less than 0.1% or from 0.1% to 1% of the market share), the ceteris paribus assumption should be applied, whereas, the effects with large potential consequences (accounting for more than 1%) should be approached by CLCA with mutatis mutandis assumption [36]. Although the criteria were not adequately convincing, they were the initial efforts of how to deal with system expansion in CLCA, based on quantitative economic value. Due to the small economic size of the product system, the potential effects of any change on the product system is limited in physical changes; namely, changes of quantity of environmental impacts without affecting the economic systems. If the effects are large in scale, there will be changes in economic interactions in other relevant sectors. In this case, the expansion of system boundary is necessary to accurately quantify the impact. Frischknecht and Stucki concluded that CLCA with mutatis mutandis assumption, therefore, would be relevant for quantifying impacts of important decisions on the investigated object with a relatively large economic size, for example governmental policies or strategic international organization decisions [36].

Examples of the mutatis mutandis assumption can be found in several studies [23-34, 40-45] (Refer to Table 2). In these studies, the marginal products or technologies were identified, with determined affected scales. The changes in the affected products were determined by reviewing literatures [24, 27, 30, 34, 40], and running economic models [23, 25-34, 40-45].

Specifically, Moora and Lahtvee quantified the global warming potential of waste incinerated electricity in Baltic region by 2020 [24]. The authors calculated the net emissions of energy generation from waste incineration (E_net) by subtracting the emissions of waste for electricity generation (E_waste) from emissions caused by alternative energy production (E_alternative), with following equation: E_net = E_waste – E_alternative. The alternative energy productions were identified as the marginal electricity in short term (electricity from natural gas, coal fired electricity and oil shale) and in long term (electricity from natural gas, coal fired electricity, biomass based and wind power) [24]. One limitation of the study is that the substitution ratio of marginal electricity with waste incinerated electricity is assumed as 1:1, which makes it in-between ceteris paribus and mutatis mutandis assumptions as the investigated product system equals to the identified marginal products.

Mathiesen et al. assessed the environmental impacts of 10% increase of waste incineration for energy production. The affected energy technologies were various among different approaches. By looking at publications on historical and future energy system and existing CLCAs, the identified affected product was coal-fired energy. Meanwhile, with the application of an economic model, e.g., EnergyPLAN, the identified affected products were mostly energy from natural gas and a small amount of coal [27]. The author concluded that with CLCA it is impossible to identify one “single” marginal technology and one “single” marginal technology would not accurately reflect the actual operation and interactions of the power system.

Gibon et al. conducted the LCA of human health and ecological impacts of the global low-carbon electricity due to carbon emission reduction policy. The low-carbon electricity scenario was determined by analysing the existing and future power generation mixes and the technology performance from International Energy Agency reports. The improvement in material production was taken from the New Energy Externalities Development for Sustainability project. The changes in the future power system structure and fuel consumption due to the increased adoption of clean power technology was identified based on experts’ opinions [30]. The life cycle inventories of these clean power technologies, with improvements in material production and power consumption, were modelled from EXIOBASE, a multi-regional input output model. It was identified that, thanks to the lower impacts on climate change, renewable energy and nuclear power cause lower human health impacts than coal or gas power. However, in term of ecological impact, the biopower had the similar ecosystem quality damage points (species.yr/kWh) as the coal power. The land occupation of biopower causes damage to the ecosystem quality at the same magnitude as the climate change impact of coal power [30].

Vandepaer et al. quantified potential environmental benefits and drawbacks of batteries integration into the future Swiss electricity supply system. The marginal electricity supply mixes were extracted from literatures of Kannan and Turton for electricity generation scenarios, publications of European Commission and International Energy Agency for electricity imports. The projections about the battery performances and the recycling process were assumed based on relevant literatures [34].

Dandres et al. assessed the environmental impacts due to European energy generation and perturbation of world economy. Different energy development scenarios of the business-as-usual (BAU) and future renewable technology mixes were taken from peer-reviewed publications of Mantzos et al. [40].

A common feature of Mathiesen et al., Gibon et al., Vandepaer et al. and Dandres et al.’s studies is that they made the assumption on the affected product by reviewing existing literatures, and combining with economic modelling, for examples EnergyPLAN [27], EXIOBASE [30], Swiss TIMES Energy Model (STEM) [34], and Global Trade Analysis Project (GTAP) [40].

Table 2. System boundary - product systems and affected products of some reviewed papers

Other authors used economic models to determine marginal technologies such as Balmorel [23], EU Electricity Market Model (E2M2) [25], JRC-EU-TIMES [26], EnergyPLAN [28], ETEM and LUXGEM [29], THEMIS and EXIOBASE [30], TIMES (ETM-UCL) and Environmentally Extended Input-Output (EEIO) [31], Unit commitment model [32], Emissions Reduction Cooperation Model [33], Sequential Quadratic Programming algorithm [41], MARKAL [42], Network Impact Assessment Model [43], Agent Based Modelling (ABM) [44] and System Dynamics (SD) [45]. The results of the studies that applied economic models will be presented in the following parts.

Under the mutatis mutandis assumption, the number of identified affected products does not limit in one product, but a combination of several products (Refer to Table 2). Moreover, the scales of change are different among affected products. In other word, a change in the investigated product system causes necessary changes in various affected products. The review on system boundary and assumptions of existing literature indicated that the mutatis mutandis assumption was preferred for CLCA in power sector, and in order to determine the set of effected products to the extent of change, the application of economic modelling is useful.

3.2 Socio-economic interactions

According to Earles and Halog, Zamagni et al., Weidema, CLCA included the economic relations of the product system in forms of market mechanism or economic-based causal relationship [15, 18, 21]. The most common way to model the economic relationship is combining an economic modelling tool and LCA. The applicable economic models are either partial equilibrium (PE), general equilibrium (GE), input output (IO) or dynamic models (DM), which were also common in CLCA studies in the energy sector in general [46]. In power sector, two thirds of reviewed papers applied economic models to simulate the economic interactions between power sector and other sectors. The pathway for integrating these models into CLCA was running the models to obtain scenarios with changes in affected sectors and identify affected products/technologies. These scenarios and data on affected products/technologies would be used for running CLCA.

Among reviewed papers, there were ten studies using PE, two studies applying GE, five studies applying IO, and two studies applying DM. More interestingly, it was common that some studies apply several types of economic models, for example PE in combination with GE, and PE in combination with IO. The applications of PE, GE, and IO had the advantage of available data and energy models. Meanwhile, the DM was limited in terms of availability of data and modelling tools, but worked well with socio-economic data [46]. Table 3 presents the reviewed CLCA case studies and their applicable models.

The applications of several economic models, e.g., IO/ PE/GE in CLCA provided details of the economic causal relationship [47]. PE had the advantage of conveying a specific view of one economic sector, while GE delivered a comprehensive view of the product system in relations with the economy [46]. IO can be applied to either one economic sector, e.g., EnergyPLAN for modelling power sector, or used for studying the whole economy, e.g., World Input Output Database [48].

Table 3. Applicable tools in reviewed papers

Hammond and O'Grady presented three pathways to a low carbon power sector of UK by 2050, through coal phase out, and technological innovations in CCS, and combined heat and power. These pathways were developed through Stakeholders workshop; Quantitative research; and Interdisciplinary workshop. The authors mentioned of applying economic models and interdisciplinary assessment in “quantitative research” step; however, it is not clear which models have been used [47].

Igos et al. applied LUXGEM and ETEM models to evaluate the economic impacts of policy decisions on energy commodity demand in Luxembourg by 2030 and identify least cost technologies to meet that energy demand [29]. LUXGEM is a GE model, which maximizes one representative household’s utility with budget constraint. It includes 20 commodities and 4 production factors of capital, labour, energy, materials. The results obtained with LUXGEM were economic interactions, including energy commodity demand. The economic drivers and energy commodity demand from LUXGEM were used to run ETEM. ETEM is a PE model, which minimizes the net present value of all costs of both supply side and demand side of the energy system. ETEM provides technology detail of energy sectors with 24 final demand commodities. The simulations of LUXGEM and ETEM were repetitively conducted to obtain the same energy consumption. It was identified that the contribution of other economic sectors to the national environmental profile, except for energy sector, were quite similar. Moreover, most of the environmental impacts originated from imported commodities [29].

McDowall et al. combined ETM-UCL and EEIO to quantify indirect carbon emission of renewable energy technologies of EU energy system [31]. ETM-UCL is a PE energy development model, which optimizes energy system cost to carbon constraints. Energy demand is exogenous, based on drivers such as GDP, population, household numbers and sectorial output. The obtained result of ETM-UCL was an optimized mix of power generation technologies. EEIO was used to obtain the direct and upstream emission factors of power generation technologies. The results on changes within the power sector were specific and interesting, which indicated the reduction of solar PV deployment when the upstream emissions were included. It was the same for the renewable energy, with 4% reduction of installed capacity by 2050 [31].

Dandres et al. applied GTAP model to predict global economic perturbation potentially caused by two different European energy policies, and CLCA to quantify environmental impacts due to these policies [40]. GTAP is a GE model, which simulates the world economy subject to changes in fuel demand, economic and technological changes. The original 57 economic sectors and 113 regions of GTAP were aggregated into 20 sectors and 14 regions. The obtained results from GTAP were comprehensive changes in production of the global economy for each region. Among economic sectors, the most impacted sector was coal extraction and power generation. Consequently, it contributed to most of the difference in the environmental impacts. Moreover, the authors pointed out that the most sensitive causal relation lied in economic revolution or the change in the GDP, rather than the change in the demand [40].

Apart from economic causal relationship, CLCA also can cover the social interrelations among the product system. In this case, the original changes were not limited to the decrease or increase of consumption and production, which are quantitative, but also include changes in social indicators. This was conducted with the application of DM such as ABM and SD.

Among the reviewed papers there were only two case studies covering the social aspects or social relationship of the power system. These studies either simulated the social agents and their impacts on the product system [44], or simulated the socio-economic interactions of the product system over its life cycle [45].

Florent and Enrico combined ABM and CLCA to model changes in vehicle private use in Luxembourg due to different mobility policies in relation with environmental impacts of battery EVs, plug-in hybrid EVs by 2020, and compared them with gasoline internal combustion vehicles and diesel internal combustion vehicles [44]. The obtained results included qualitative changes on characteristics of travel, charging patterns and auxiliary use and quantitative indicators such as number of travels, which consequently change environmental indicators such as decrease global warming, fossil depletion, acidification, ozone depletion and photochemical ozone formation; and increase in metal depletion, ionizing radiations, marine eutrophication and particulate matter formation [44].

In Onat at el.’s study, CLCA based on SD was applied to model impacts of three alternative vehicles including internal hybrid, plug-in hybrid and battery EV by 2050 and compared them with internal combustion vehicle [45]. The increase in the utilization number of EVs caused sustainable impacts on three pillars of sustainability (environment, economy and society), including carbon dioxide emissions, particulate matter formation, photochemical oxidant formation (for environmental aspect), vehicle ownership cost, contribution to GDP (for economic aspect), employment generation, and the human health impacts (for social aspect). With CLCA approach, it was identified that EVs were expected to be the best alternative in long-term for reducing human health impacts and air pollution from transportation. Meanwhile, the result based on average value indicated that plug-in hybrid vehicles had the largest potential GHG emission reductions [45].

In the context of GHG policy intervention, more and more renewable energy sources will be integrated into the power system. The impacts of these integrations will not limit in the technical changes of the power system, and thereby its relevant environmental consequences, both positive and negative, but also cause economic and social impacts. As a result, the development of CLCA methodology with the consideration of socio-economic interactions is useful for comprehensively and precisely assessing impacts of the future power system.

3.3 Indirect environmental impacts of the power sector

By expansion of system boundary and inclusion of socio-economic interactions, CLCA demonstrates its ability in quantifying indirect life cycle environmental impacts of the power sector. When the indirect environmental impacts are included, the total environmental impact points either increase or reduce. In other words, the investigated power sector becomes cleaner (increased emission reduction), less clean (decreased emission reduction), more polluted (increased emissions) or less polluted (decreased emissions). In the analysed literature, the scales of change range from 5% to 200% [50]. The difference between the life cycle environmental impacts with and without indirect impacts quantified with CLCA approach in the power sector is presented in Table 4.

Frischknecht and Stucki used attributional, decisional (consequential at micro-economic level) and consequential approaches to quantify the environmental impacts of French and EU electricity supply [37]. The attributional life cycle inventory was taken from Ecoinvent database, the most common life cycle inventory database. The decisional life cycle inventory was based on EurElectric (a publication on power statistics in European countries), other energy publications and expert opinions. The authors identified that there was a difference among the obtained results. In the French electricity supply mix, the GHG emissions calculated with attributional and decisional approach varied by 129.6%, at 98 gCO₂e per kWh and 225 gCO₂e per kWh, respectively. However, the volumes of high radioactive waste reduced from 11 with attributional approach to 3.8 mm³ per kWh with decisional approach [36]. In EU electricity supply mix, both environmental indicators of GHG emissions and high radioactive waste reduced when decisional approach was applied, from 554 to 473 gCO₂e per kWh and from 3.5 and 3.4 mm³ per kWh, respectively [37]. The decreases of GHG emissions and high radioactive waste of the EU electricity are due to the differences between the average electricity mix and the decisional electricity mix, in which the latter has lower GHG emission thanks to the phasing-out of electricity from hard coal and lignite [37].

Pehnt et al. studied the increase in offshore wind power and its benefits in GHG emission reduction in Germany [25]. The change in offshore wind power capacity induced a change in the power mix, operation of power system, expansion and reinforcement of the grid. In terms of power mix, the increased offshore wind power substituted the thermal power from coal, lignite and gas. Thanks to the substitution effect, the specific carbon reductions per kWh offshore power in 2020 would amount to 914 and 646 gCO₂e, in the low and high carbon certificate price scenarios, respectively [25]. At the same time, the operation of thermal power plants would be affected, adding 70 and 18 gCO₂e of emissions per kWh of off-shore wind power. The emission from wind induced grid expansion and reinforcement is 22 gCO₂e per kWh. When the emissions from all processes, including construction, operation and disposal of the wind energy park, wind-influenced grid expansion, carbon reductions due to thermal power substitution and GHG emissions from altered power plant operation were added up, the net carbon reductions were 822 gCO₂e per kWh and 606 gCO₂e per kWh [25].

Blanco et al. conducted an ex-post LCA analysis of results from a PE based energy system model called JRC - EU – TIMES and estimated the 18 environmental impact indicators across five energy-related sectors of power, heating, industry, transport and supply in scenarios that achieved 80-95% GHG emission reductions by 2050 in EU28+ countries [26]. The investigated environmental impacts included climate change, fossil depletion, metal depletion, water depletion, ozone depletion, ionizing radiation, particulate matter formation, photochemical oxidant formation, terrestrial acidification, freshwater eutrophication, marine eutrophication, marine ecotoxicity, freshwater ecotoxicity, terrestrial ecotoxicity, human toxicity, natural land transformation, urban land occupation and land occupation. The results showed that the indirect GHG emissions were as large as the direct one for 80% reduction target. For 95% reduction target, the indirect GHG emissions were three times larger than the direct one [26]. As the emission reduction target increases from 80% to 95%, most categories' impacts decreased by 20-40%, except for toxicity which was higher at 35-100% [26].

McDowall et al. conducted a similar study on renewable energy technologies of EU energy system. In the study, the indirect emissions contributed to less than 10% of the total emissions of power sector in EU by 2050 [31], which was a small part, especially compared to the result of Blanco et al.’s study. It should be noted that there is a difference in investigated scopes of these studies. While Blanco et al.’s study covered five energy-related sectors, the indirect emissions in McDowall et al.’s study included those from energy equipment and construction [26, 31]. Even though the indirect emission of the power sector is small in McDowall et al.’s study, its inclusion into the optimization model made the renewable power less attractive than it was expected [31].

Table 4. Direct and indirect environmental impacts of some reviewed CLCA papers

* These numbers were estimated by the authors based on Figure 3 of Vandepaer et al. [34].

Igos et al. assessed the impacts on human health, ecosystem and resources of two energy policies: Business As Usual (BAU) and Greenhouse Gas (GHG). In the BAU policy, the drivers for change were socio-economic drivers, including growth rates of economic sectors, population and GDP. In the GHG policy, apart from socio-economic drivers, a reduction target of 2.5% was also considered. The environmental impacts included impacts from energy production and energy import as well as indirect impacts due to economic changes. The contribution of indirect impacts was up to 50% in all three categories of human health, ecosystem and resources. The environmental impacts in the GHG policy were 2-3% lower than those in the BAU policy. This difference mainly and directly came from energy sector. The contribution of other economic sectors such as agriculture, construction, etc. to this 2-3% difference was less than 0.1% [29].

Raugei et al. conducted a CLCA on increased uptake of solar PV in the UK grid. The increase of solar PV capacity influenced the generation mix as well as the grid development, and consequently global warming potential of solar PV [32]. The authors identified that there was a small difference in the GHG emissions from the increased solar PV deployment, which originated from background stages of solar PV and changes in the generation mix [30]. Consequently, any change in the solar PV deployment had no considerable additional emissions of the UK on-grid power, at around $\pm$ 2% [32].

Vandepaer et al. quantified the environmental impacts of inclusion of battery into the Swiss power system by 2030. In the current policy scenario, the inclusion of battery caused the displaced electricity mix, which was dominated by natural gas combined cycle units [34]. The inclusion of batteries generated environmental benefits in 12 of the 16 impact categories, including climate change, ozone depletion, particulate matter, ionizing radiation, photochemical ozone formation, acidification, terrestrial eutrophication, freshwater eutrophication, marine eutrophication, land use and water resource depletion. In low carbon scenario, marginal electricity generation being displaced due to the inclusion of batteries mostly came from geothermal and hydropower which already had reduced environmental impacts. Therefore, the environmental benefits due to the inclusion of battery are reduced as compared to those of current policy scenario [34]. Specifically, in the current policy scenario, the emission reductions of lithium metal polymer and Li-ion battery are 443 and 439 kgCO₂e/MWh, respectively. In the low carbon scenario, these numbers are less than 200 kgCO₂e/MWh for both types of batteries [34].

Dandres et al. assessed the environmental impacts of EU electricity and heat generation in two EU energy policies, namely baseline and bioenergy, in consideration with and without global economic development. The quantified impacts included direct impacts of increased energy generation and indirect impacts due to change in the global economic activities served for increased energy generation in the EU. It was indicated that potential indirect impacts were higher than direct impacts, with impacts occurring inside the EU border only accounting for 5.5% of the total global potential impacts. Interestingly, indirect impacts of increased energy in bioenergy policy were considerably higher than those in baseline policy [40]. Bioenergy policy, which harnesses more renewable energy compared to baseline policy, should be cleaner. However, the obtained results indicated that it causes more environmental impacts due to the indirect consequences of bioenergy on global economic activities [40].

There is a link between the environmental and economic aspects of power sector. Any increase or decrease in the indirect environmental impacts of the power system will affect the cost for the system. Algunaibet et al. quantified the life cycle indirect cost of electricity generation in the US and pointed out that other indirect environmental impacts of power sector need to be considered apart from direct GHG emissions. In the study, the costs of electricity were minimized with constraints on demand, generation potential and capacity factor, while achieving a particular target on emission. These costs included levelized cost of electricity (direct cost) and costs to endpoint life cycle indicators including human health, ecosystem diversity and resource availability (indirect cost, externalities) [31]. It was found that by meeting the emission reduction of Paris Agreement, the indirect costs of electricity generation could be reduced up to 63% [31]. In contrast, the withdrawal from Paris Agreement would cause a cost up to 1103 $\pm$  206 billion USD 2013 in BAU scenario by 2030 [31]. When both direct and indirect costs were optimized, the total cost for the energy system would be 373 $\pm$ 164 billion USD 2013 in 2030 [31].

Elzein et al. assessed the GHG emission and operating cost of electricity generation of Normandy grid with the inclusion of energy storage systems. The inclusion of energy storage systems altered the generation of thermal and nuclear power plant, consequently, reduced GHG emissions by 53%. At the same time, the operating cost reduced by 28% as compared to the base case of historic power generation without energy storage systems [38].