This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 147.125.96.230

This content was downloaded on 02/06/2016 at 14:56

Please note that terms and conditions apply.

Energy sector water use implications of a 2 °C climate policy

View the table of contents for this issue, or go to the journal homepage for more

2016 Environ. Res. Lett. 11 034011

(http://iopscience.iop.org/1748-9326/11/3/034011)

Home Search Collections Journals About Contact us My IOPscience

Environ. Res. Lett. 11 (2016) 034011 doi:10.1088/1748-9326/11/3/034011

LETTER

Energy sector water use implications of a 2°C climate policy

Oliver Fricko1, SimonCParkinson1,2, Nils Johnson1,Manfred Strubegger1,Michelle TH vanVliet1,3 andKeywanRiahi1,4

1 International Institute for Applied SystemsAnalysis, Austria2 Institute for Integrated Energy Systems, University of Victoria, Canada3 Earth SystemScience,WageningenUniversity, TheNetherlands4 GrazUniversity of Technology, Austria

E-mail: fricko@iiasa.ac.at

Keywords: energy, water, climate changemitigation, thermal water pollution

Supplementarymaterial for this article is available online

AbstractQuantifyingwater implications of energy transitions is important for assessing long-term freshwatersustainability since large volumes of water are currently used throughout the energy sector. In thispaper, we assess direct global energy sector water use and thermal water pollution across a broad rangeof energy system transformation pathways to assess water impacts of a 2 °C climate policy. A globalintegrated assessmentmodel is equippedwith the capabilities to account for thewater impacts oftechnologies located throughout the energy supply chain. Themodel framework is applied across abroad range of 2 °C scenarios to highlight long-termwater impact uncertainties over the 21st century.Wefind that water implications vary significantly across scenarios, and that adaptation in power plantcooling technology can considerably reduce global freshwater withdrawals and thermal pollution.Global freshwater consumption increases across all of the investigated 2 °C scenarios as a result ofrapidly expanding electricity demand in developing regions and the prevalence of freshwater-cooledthermal power generation. Reducing energy demand emerges as a robust strategy forwaterconservation, and enables increased technological flexibility on the supply side to fulfill ambitiousclimate objectives. The results underscore the importance of an integrated approachwhen developingwater, energy, and climate policy, especially in regions where rapid growth in both energy andwaterdemands is anticipated.

1. Introduction

Access to water in the energy sector is crucial forresource extraction, fuel processing, and electricpower generation (Mielke et al 2010, Macknicket al 2012a). The volume of water used directly in theenergy sector is considerable, representing approxi-mately 15% of global freshwater withdrawals in 2010(Flörke et al 2013, IEA 2012). The scale of energy sectorwater demand exerts pressure on the global hydro-logical cycle (Döll et al 2012), and could causeallocation conflicts with the agricultural, manufactur-ing and domestic sectors in areas facing watershortages (Wimmer et al 2015). Future energy systemtransformations are thus important to consider when

assessing the long-term sustainability of waterresources.

Limiting global mean temperature change overpre-industrial levels to 2 °C will require a completetransformation of the global energy system (Riahiet al 2012, Kriegler et al 2014). Although the 2 °Ctemperature target is clearly framed, the specific char-acteristics of the required energy system transitionmay differ significantly (Riahi et al 2012, Kriegleret al 2014). As water impacts vary across energy tech-nology options (Mielke et al 2010, Macknicket al 2012a), quantification of the associated water useuncertainties for a large range of energy transition sce-narios is needed to accurately assess future risks towater resources.

OPEN ACCESS

RECEIVED

6October 2015

REVISED

18 January 2016

ACCEPTED FOR PUBLICATION

28 January 2016

PUBLISHED

4March 2016

Original content from thisworkmay be used underthe terms of the CreativeCommonsAttribution 3.0licence.

Any further distribution ofthis workmustmaintainattribution to theauthor(s) and the title ofthework, journal citationandDOI.

© 2016 IOPPublishing Ltd

The aim of this paper is to quantify direct energysector water impacts globally across a broad range ofenergy system transformation pathways under a 2 °Cclimate policy. Previous analysis at both the regionaland global-scale focused mainly on low-carbon elec-tricity generation pathways (Macknick et al 2012b,Davies et al 2013, Kyle et al 2013, Webster et al 2013,Bouckaert et al 2014, Byers et al 2014, Cameronet al 2014). These studies highlight the sensitivity ofwater withdrawal and consumption to technologychoice, and the potential benefits of deployment stra-tegies that simultaneously reduce electricity sectorwater use and emissions. In this paper, we enhance aglobal integrated assessment model with the cap-abilities to track water withdrawal and consumptionacross the energy supply chain (extraction, fuel pro-cessing, electricity generation and heat production).Moreover, we perform a prospective analysis of ther-mal water pollution from thermoelectric powerplants, which poses a threat to aquatic ecosystems(Chuang et al 2009, Stewart et al 2013), and to date, hasnot been assessed globally. We further quantify sensi-tivities arising from shifts in power sector coolingtechnologies and energy demands. The analysis pro-vides important insight into how energy transitionsconsistent with a 2 °C climate policy influence regio-nal and global water trends, and the associateduncertainties.

The paper proceeds as follows. In section 2 wedescribe the energy transformation pathways investi-gated and the methodology for identifying the waterimpacts (withdrawal, consumption and thermal pol-lution) of different energy technologies. Results fromthe scenario analysis are presented in section 3, whichfocus on global and regional outcomes obtained acrossthe scenarios and water impact uncertainties due tomodel parameterization. A discussion of results andmain conclusions are provided in section 4.

2.Methods

2.1. Energy system transformation pathwaysWe quantify the water impact of energy systemtransformation pathways developed in the globalenergy assessment (GEA) (Riahi et al 2012). The GEApathways were designed to describe transformativechanges toward amore sustainable future, and includea 2 °C climate policy. We chose the GEA scenarios forour analysis because the broad range of energytransitions covered by the scenario space provides anideal platform to assess uncertainties in future waterdemand stemming from technology choices made inthe energy sector.

Each GEA energy transformation pathway isdefined by a unique combination of branching pointssummarized in figure 1. Three pathway groups areinitially distinguished by level of energy demand:GEA-Efficiency, which emphasizes demand-side and

efficiency improvements at a relatively low demand-level; GEA-Supply, which emphasizes supply sidetransformation at relatively high energy demand; andGEA-Mix, which emphasizes regional diversity at anintermediate level of energy demand. Further scenariobranching points account for differences in the pro-gression of the transportation system. Advancedtransportation scenarios involve accelerated diffusionof electric and hydrogen vehicle technologies, while aconventional transport pathway entails continued useof liquid fuels. The pathway groups contain furtherscenarios that explore alternative transformations onthe supply side. These scenarios restrict developmentof particular technology options to test the relativeimportance of technology availability in fulfilling the2 °Cclimate policy objective.

The GEA energy transformation pathwaysexplored in the analysis are generated with the globalintegrated assessment model for energy supply sys-tems and their general environmental (MESSAGE)impact: a linear systems-engineering optimizationmodel that solves for the energy technology portfolioand land-based mitigation measures in 11 globalregions over a planning horizon spanning the 21stcentury (supplementary information, section S1)(Strubegger et al 2004, Riahi et al 2007, 2012). MES-SAGE is used in conjunction with model for green-house gas induced climate change (MAGICC) forcalculating internally consistent scenarios for limiting21st century global mean temperature change to 2 °Cwith a probability exceeding 50% (Wigley andRaper 2001, Rogelj et al 2013a). Some scenariobranching points lead to infeasibility in terms ofreaching a 2 °C temperature target, with a total of 41scenarios found to fulfill the climate policy objective.This scenario space spans a much larger range thanprevious global assessments (Davies et al 2013, Kyleet al 2013), and thus the analysis here provides amore detailed view of potential water impact uncer-tainties arising from different energy transformationpathways.

2.2.Water impact assessmentWe adapt the MESSAGE IAM framework to calculatethe energy sector water impacts of the GEA transfor-mation pathways. The majority of energy sector fresh-water withdrawal occurs in the steam-cycle andcooling systems of thermoelectric generation(IEA 2012, Flörke et al 2013), and we assess water useby thermoelectric power plants included in the IAM asa function of the thermal conversion efficiency. Thefollowing equation is used to express water withdrawalor consumption intensity i (e.g., m3 kWh−1 net poweroutput) as a function of heat-rate (how efficiently theplant converts heat to electricity), and cooling systemtype (Delgado andHerzog 2012):

i , 1· ( ) ( )a b d= - +

2

Environ. Res. Lett. 11 (2016) 034011

where ò represents the heat-rate (kWh heat/kWh netpower output), α represents how efficiently the cool-ing technology utilizes water (m3 kWh−1 heat), β

represents other heat outputs (heat content of electri-city and other heat losses such as with flue gases; kWhheat/kWh net power output), and δ represents waterrequirements other than for cooling (m3 kWh−1 netpower output). The parameter α varies across coolingtechnologies, and is calculated using a physically-based approach that relates the amount of waterwithdrawn or consumed to how the cooling systemcirculates water and the allowed increase in effluenttemperature (Delgado and Herzog 2012). Once-through and closed-loop cooling technologies aredistinguished in this context. Once-through coolingtechnology, as the name suggests, involve passingwater through the cooling system once, and thenreturning the water to its source. Conversely, closed-loop systems re-circulate water that is withdrawn. Thewater source (fresh or saline) is further distinguishedacross technologies. We also consider air-cooledsystems, which provide an opportunity to reduceenergy system reliance onwater.

This choice of model formulation enables con-sistent representation of water use across power planttypes and incorporates water impacts of heat-rateimprovements due to anticipated long-term technolo-gical change. Moreover, the approach enables analysisof thermal water pollution from once-through cooledthermal power plants by allowing quantification of the

heat energy embodied in cooling system effluents.Specifically, we treat all heat energy remaining afterelectricity conversion and air emissions as thermalwater pollution (i.e., b- for once-through sys-tems).We parameterize the thermal power plant waterimpact model following the analysis in Delgado andHerzog (2012). The calculated water intensities areprovided as online supplementary data, and werecompared with literature estimates and good agree-ment was found across sources and technologies (sup-plementary information, figure S2.1). The calculatedwithdrawal and consumption intensities are multi-plied by the optimized power plant activity in eachmodel year to estimate the regional water demands. Allcooling technology parameters other than the powerplant heat-rate arefixed across the simulation horizon.

Two cooling technology scenarios are dis-tinguished in which future shifts in cooling technologyshares are exogenously defined (figure 1). The alter-native cooling technology scenarios are used todemonstrate the potential sensitivity of the results tocooling technology assumptions, and to explore trade-offs between climate and water sustainability objec-tives. In the first scenario, entitled ‘baseline coolingtechnologies’, the future cooling technology distribu-tion for each generation technology is initially fixedacross the simulation horizon and in each region basedon the current global shares assessed by Davies et al(2013) (supplementarymaterial, section S3). Although

Figure 1. Scenario components of the 2 °Cpathways considered. A single pathway is defined by combining one component from eachcolumn. A total of 41 scenarios are found to fulfill the 2 °Cobjective (Riahi et al 2012), resulting in 82water impact scenarios. Areference case representing a situationwhere no climate policy is implemented (i.e., unconstrained emissions) is also analyzed, andcorresponds with theGEA-Mix demand-level, conventional transport development, and full portfolio supply technology availability.

3

Environ. Res. Lett. 11 (2016) 034011

the distributions are fixed in each region for each tech-nology, investment into alternative technology portfo-lios can change the distribution of coolingtechnologies in future years when aggregated to theregional-scale.

In the second scenario, entitled ‘adapt coolingtechnologies’, opportunities to address freshwaterchallenges in the energy sector through the adoptionof alternative cooling technologies and water sourcesare examined. Specifically, once-through freshwatercooling systems are phased out over the 2040 to 2060period and replaced with a combination of air and sea-water cooling technologies. Closed-loop cooling tech-nologies consume more water than once-throughtechnologies (supplementary information, sectionS2), making a transition towards seawater and air-cooled technologies essential to improving consump-tion intensity; however, it is important to note that theconsumption volume for closed-loop systems is about1% of the withdrawal volume at equivalent once-through systems. The phase-out of once-through sys-tems and shift towards air cooling and seawater tech-nology is not meant to represent the optimal strategy.Instead, we use the transition towards air cooling tohighlight potential efficiency tradeoffs (Turchiet al 2010, Zhai et al 2011).Moreover, we use the trans-ition to seawater cooling technology to demonstratethe tradeoffs with marine thermal pollution. The pol-icy is simulated by forcing the regional shares of once-through freshwater systems to zero along a linear tra-jectory, with 90% of the affected capacity transitionedto air-cooling and 10% to seawater cooling. The regio-nal shares of closed-loop freshwater systems remainfixed for each technology over the simulation period.Less capacity is transitioned to seawater coolingbecause these systems thermally pollute the marineenvironment. The timeframe for the transition isselected because it aligns with the expected ramp-up incapacity investment resulting from retirements ofexisting and planned infrastructure. Average efficiencylosses due to air-cooling are estimated for each tech-nology (supplementary information section S3). Sea-water cooling also potentially requires extra energy forwater treatment and maintenance, although theseeffects are excluded due to lack of suitable data.

Water impacts of other energy technologies inclu-ded in the analysis (non-thermal power plants and fuelprocessing/resource extraction technologies) are cal-culated with operational water coefficients. Theseparameters describe the average amount of waterwithdrawn or consumed per unit of technology out-put, and are estimated based on a comprehensivereview of previous studies (supplementary informa-tion, section S2). The implemented coefficients areincluded as online Supplementary Data. Regionalfreshwater consumption and withdrawal from thesetechnologies are then calculated by multiplying thecorresponding water coefficient by the optimizedtechnology activity in each model year. Literature

estimates vary considerably due to a combination oftechnology vintages, local climate conditions, systemconfigurations, and reporting methods (Macknicket al 2012a). We explore these uncertainties by con-ducting a sensitivity analysis across the range of repor-ted values.

2.3. LimitationsOur analysis excludes differences in cooling technol-ogy costs, which are likely to impact the economics ofclimate change mitigation pathways (Websteret al 2013, Tidwell et al 2014). Constraints on theavailability and quality of water resources (e.g., watertemperature) pose risks to energy supply reliabilityand vary significantly across the macro-regions understudy (vanVliet et al 2013), but are also excluded in themodel. These characteristics are difficult to emulate ina global optimization model due to water constraintsoccurring at relatively small temporal and spatial scales(Sun et al 2015). Furthermore, our analysis does notinclude the potential irrigation water used for bioe-nergy feedstock cultivation (Gerbens-Leeneset al 2009). However, the bioenergy included in ourassessment considers a number of different sustain-ability constraints that promote use of short-termwoody crops grown in areas not requiring irrigation(Riahi et al 2007, van Vuuren et al 2009). Theprecipitation incorporated into the biomass could besignificant (Gerbens-Leenes et al 2009), and increasedland area needed to cultivate these crops could pushother irrigated crops into areas requiring increasedirrigation, but these effects occurring upstream fromthe energy sector are excluded.

3. Results

3.1. Influence of technological pathway on globalwater trendsFigure 2 depicts the global water withdrawal, con-sumption and thermal pollution associated with thetwo different cooling technology scenarios across aselected set of 2 °C energy transformation pathways(GEA-Mix), a reference scenario (no climate policy),and uncertainty ranges across all 2 °C scenarios tested.Despite increasing electricity demand, global waterwithdrawals remain relatively steady out to 2040 as aresult of the improvements in withdrawal intensitythat accompany a shift from steam- to combined-cyclefossil-based power generation (Davies et al 2013, Kyleet al 2013, Tidwell et al 2014). After 2040, withdrawalsdiverge substantially. In the ‘baseline cooling technol-ogies’ scenario, we find that cases consistent with 2 °Cresult in end-of-century withdrawals changingbetween −10% and 611% relative to base year (2000)conditions (figure 2(A)). Once-through cooled ther-mal power generation remains the dominant source ofwater withdrawal across the mitigation scenarios dueto the current prevalence of the technologies in

4

Environ. Res. Lett. 11 (2016) 034011

regional shares. The proportional relationship derivedbetween once-through withdrawals and thermal pol-lution means that similar trends for thermal pollutionare obtained (increases between −54% and 638%

relative to base year conditions). The wide range inresults is a consequence of the large differences inwithdrawal intensity observed across the low-carbonenergy technologies (supplementary information,

Figure 2.Global water impacts across the 2 °Cand reference scenarios for the two thermal power plant cooling technology cases: (A)baseline cooling technologies; and (B) adapt cooling technologies. Individual scenario results are illustrated for a subset of climatechangemitigation and reference scenarios with intermediate energy demand (GEA-Mix). The full range of water impacts associatedwith all technology scenarios are illustrated for each energy demand assumption (GEA-Efficiency, GEA-Mix, andGEA-Supply). Theadditional range resulting from themaximumandminimum reportedwater intensity coefficients are indicated by gray lines.

5

Environ. Res. Lett. 11 (2016) 034011

figure S2.1), and the various combinations of technol-ogies commensurate with 2 °C. For example, nuclear,carbon capture and storage (CCS), and concentratingsolar power (CSP) technologies are associated withvery low carbon emissions, but can require significantwater for cooling or processing. Conversely, low-carbon technologies such as wind and solar PV havevery lowwater requirements.

In the ‘adapt cooling technologies’ scenario (i.e.,phase-out of once-through freshwater systems overthe 2040 to 2060 period), freshwater withdrawalsdecrease across all scenarios after 2040 and much lessuncertainty is observed (−3% to −63% of base yearlevels in 2100) (figure 2(B)). Thermal pollution is lessaffected than freshwater withdrawal (−89% to 312%of base year levels in 2100), as the capacity transferredto seawater cooling thermally pollutes the marineenvironment.

Different patterns are observed in the results forglobal water consumption, which show increasingrequirements across all scenarios (215%–747%increase relative to the base-year). Freshwater con-sumption levels reduce slightly in the ‘adapt coolingtechnology’ scenario compared to the baseline case(146%–543% increase relative to base year levels) dueto the increased use of air and seawater cooling. How-ever, even in the adaptation scenarios, freshwater con-sumption continues to expand as closed-loop coolingsystems continue to be used to meet the growing elec-tricity demand. More broadly, the growth in fresh-water consumption follows from the increasinglyimportant role of electricity in climate change mitiga-tion scenarios. Electrification of end-use is a commonmitigation theme in the GEA scenarios, leading toincreased global electricity demand (Riahi et al 2012).The prevalence of freshwater-cooled CCS, nuclear,and CSP, as well as hydropower across the differentscenarios makes electricity more water consumption-intensive than conventional liquid and solid fuels atthe final energy-level. This would suggest thatalthough electrification is likely to improve energyintensity and enable access to low-cost, low-carbonenergy technologies (Riahi et al 2012), there are con-current increases in water consumption anticipated inthe GEA scenarios. Less variability is observed acrossconsumption scenarios as compared to withdrawalbecause the water consumption intensity varies lessamong the low-carbon electricity generation optionsprevalent in theGEA scenarios (i.e., CCS, CSP, nuclearand hydropower) (supplementary information,figure S2.1).

The majority of mitigation pathways are found toexceed the projected water demand (withdrawal andconsumption) and thermal pollution under the refer-ence scenario (no climate policy). The reference caseuses less water due to expansion of combined-cyclenatural gas power plants when climate constraints areexcluded. Mitigation scenarios limiting expansion ofnuclear and CCS technologies result in the lowest

water impacts, but also some of the highest expectedmitigation costs (Riahi et al 2012). Hydropower andCSP encompass most of the remaining consumptionand thermal pollution in the scenarios involving lim-ited CCS and nuclear. Evaporation associated withstoring water in hydropower reservoirs constitutes thelargest component of base year water consumption(59%), with expanding nuclear, CCS, and CSP genera-tion dominating the increased global requirementsprojected in future years. Water consumed during theextraction and processing of coal and oil resourcesdecreases significantly across the mitigation scenariosthat limit CCS deployment. Freshwater consumed fornatural gas extraction and processing expands to levelscurrently seen for coal resources by mid-century(reflecting a switch from coal to gas), and reducesthereafter to enable emission levels consistent with2 °C. Extraction and processing water use represents arelatively small fraction of total sector demand (19%of total energy sector consumption in the base year),with the projected reductions having little impact onaggregate energy sector water use. Nevertheless, thereare concomitant water quality benefits of reduced fos-sil fuel extraction that are excluded from this analysisand important to consider in future research.

Further depicted in figure 2 is the uncertaintyrange obtained from varying the technology-levelwater withdrawal and consumption coefficients esti-mated from the reviewed performance data (onlinesupplementary data). For withdrawals, the uncertaintydue to technology parameterization is found to berelatively minor in comparison to the range obtainedacross the GEA scenarios. In contrast, the uncertain-ties associated with water consumption coefficientsare much larger. The consumption of hydropowergeneration is the main contributor to this uncertaintyas it is difficult to accurately assess and attribute con-sumption to electricity production, especially formulti-purpose reservoirs where water is stored for dif-ferent economic purposes.

3.2. Cumulative impacts of cooling technologyadaptationThe ‘adapt cooling technologies’ scenario displayssignificant water benefits, but affects electricity pro-duction due to air-cooling efficiency losses. Cumula-tive impacts of the alternative cooling technologyscenario for selected energy technology portfolios aredepicted infigure 3.Wefind that transitioning towardsincreased air and sea water cooling would reduce thefull portfolio cumulative freshwater withdrawal overthe 2040 to 2100 period by 74%, thermal pollution by41%, and freshwater consumption by 19%. Tradeoffscome in the form of increased cumulative globalelectricity production requirements of 3%. Similarresults are obtained for the other selected scenarios.The limited wind/solar scenario displays the largestwater benefit from transitioning to alternative cooling

6

Environ. Res. Lett. 11 (2016) 034011

technologies due to the high-penetration of nuclearandCCS generation.

3.3. Influence of energy demandComparing the results for the ‘baseline coolingtechnologies’ scenario obtained across the differentdemand levels reveals that moving from the mid-demand (GEA-Mix) full mitigation portfolio to a lowenergy demand scenario (GEA-Efficiency) results in a45% reduction in end-of-century withdrawals, 63%reduction in thermal pollution, and 28% reduction inconsumption. Comparable reductions are obtainedacross the other technology scenarios, and suggest acritical role for end-use energy efficiency in globalwater conservation. Water inputs are required duringthe transformation of energy resources into usefulservices. As lower demands reduce the need forresource transformation, measures limiting energydemand growth (e.g., high-efficiency end-use appli-ances) are also an effective water conservationapproach.

Water consumption is particularly responsive tothe energy demand level, as depicted in figure 2. Weexplore the relationship between energy demand andwater consumption by computing the water con-sumption intensity of the energy pathways, definedhere as global water consumption divided by finalenergy demand (figure 4). The intensity of water con-sumption increases over the simulation period regard-less of the demand level, with only minor differencesobserved across the demand levels. However, it isnotable that the GEA-Efficiency scenarios display thelargest range since low demand levels affordmore flex-ibility in the supply side technologies used to mitigate

climate change (Riahi et al 2012). The wide range inwater consumption intensity results from both thescenario-specific technology restrictions and thevariability in water consumption across the technol-ogy options.

Figure 4 also indicates that the aggregate waterconsumption intensity continues to expand even inthe ‘adapt cooling technologies’ scenarios since onlyonce-through freshwater systems are converted to airand sea water cooling and thus closed-loop systemscontinue to expand. Since closed-loop systems con-sume more water than once-through technologies,relatively modest improvements to the aggregate con-sumption intensity are achieved. The reference sce-nario trajectory is also depicted, and remainsnoticeably lower than all mitigation scenarios tested.In the reference case, the transition towards com-bined-cycle power generation results in a more water-efficient system when compared to the mitigation sce-narios that involve expansion of water consumption-intensive nuclear, CSP, andCCS technology.

3.4. Regional trendsThe integrated assessment framework computesenergy sector water impacts across 11 macro-regions,and we explore the regional results in figure 5.Depicted are water withdrawal, consumption andthermal pollution results for 2010 and 2100 for the‘baseline cooling technologies’ scenario. Energy sectorwater withdrawal is projected to decrease in themajority of advanced economies (EEU, FSU, NAMand WEU) as a result of: (1) anticipated stagnation inregional energy demand; and (2) the transition awayfrom withdrawal-intensive thermal power generation

Figure 3.Cumulative impact of the ‘adapt cooling technologies’ scenario (i.e., phase-out of freshwater once-through systems)calculated from 2040 to 2100 across four of the representative technology portfolios. The percent change is calculated relative to the‘baseline cooling technologies’ scenario (i.e.,fixed cooling technology distribution for each power plant technology).

7

Environ. Res. Lett. 11 (2016) 034011

technologies (e.g., once-through cooled coal genera-tion) towards technologies with cooling system dis-tributions that include a higher proportion of closed-loop systems (e.g., combined-cycle natural gas). Mostdeveloping economies (AFR, LAM, MEA, PAS, andSAS) see an increase in withdrawals due to the rapidincrease in energy demands. Conversely, CPA andPAO achieve lower withdrawals because of the trans-ition in these regions away from once-through cooledcoal generation.

Consumption levels increase in all regions due to ahigher penetration of electrified end-uses, with

increases most prevalent in developing economiesbecause of rapidly increasing demands that accom-pany their anticipated economic advancement.Nuclear, CCS, andCSP technologies are themain con-tributors to increased water consumption and emergeas critical options for addressing the rapid demandgrowth in developing regions. In LAM, hydropowerexpansion is the primary source of increased con-sumption. Results for thermal pollution highlightregional cooling technology trends. Once-throughcooling systems are common in PAO (seawater), SAS(freshwater) and MEA (seawater), which combined

Figure 4.Water consumption intensity of the 2 °Cand reference scenarios for the two thermoelectric cooling technology cases: (A)baseline cooling technologies; and (B) adapt cooling technologies. The consumption intensity is calculated as water consumptiondivided by final energy demand (in exajoules (EJ)).

Figure 5.Regional results obtained for (a)withdrawal, (b) consumption and (c) thermal water pollution across the climate changemitigation pathways and reference scenario under the ‘baseline cooling technologies’ scenario. The depictedmagnitudes areproportional to the area of eachwedge.WEU=Western Europe; PAO=PacificOECD;NAM=NorthAmerica;MEA=MiddleEast andNorthAfrica; PAS=Pacific Asia; LAM=Latin America; SAS=SouthAsia; AFR=Sub-SaharanAfrica; CPA=CentrallyPlannedAsia; FSU=Former Soviet Union; and EEU=Eastern Europe. A full list of countries included in themacro-regions isprovided in the supplementary information (section S1).

8

Environ. Res. Lett. 11 (2016) 034011

with the regional expansion of nuclear generationacross a number of mitigation scenarios pose sig-nificantly increased risk of thermal pollution. Con-versely, Latin America has the lowest thermalpollution due to the expansion of hydropower andsolar photovoltaic (PV) technologies to meet itsrapidly expanding electricity demand. The reducedthermal pollution in other advanced economies (FSU,EEU, WEU and NAM) follows from the reductions inwithdrawals described previously.

4.Discussion and conclusions

In this paper, a global integrated assessment modelwas equippedwith the capabilities to account forwaterimpacts throughout the energy supply chain. Themodel was applied across a broad range of energytransformation pathways consistent with a 2 °C cli-mate policy to quantify long-term water impactuncertainties over the 21st century. The results of thisanalysis show that the global water implications ofenergy transitions consistent with a 2 °Cclimate policyvary significantly due to the diverse technologycombinations commensurate with a 2 °C emissionslevel and the wide range in water performance (with-drawal and consumption) observed at the technology-level. The quantified scenario uncertainties bridge amuch larger range than that seen in previous inte-grated assessments (Davies et al 2013, Kyle et al 2013).The impacts of uncertainties regarding technology-specific water withdrawal coefficients are found to berelatively minor compared with the uncertainty acrossthe range of 2 °C scenarios, while uncertaintiesassociated with water consumption coefficients have amore significant impact on the uncertainties sur-rounding energy-related water consumption.

The integrated assessment framework was furtherapplied to quantify, for the first time, global thermalwater pollution from power plants. The prevalence ofthermal power generation across the mitigation sce-narios suggests that thermal pollutionwill pose greaterrisks to aquatic ecosystems in the future. However,scenarios that phase out once-through cooling sys-tems and focus on reducing energy demand throughimproved end-use efficiency demonstrate significantpotential for reducing thermal pollution impacts. Anadditional option that was not explored in this analysisis the deployment of cooling ponds, which allow efflu-ent temperatures to decline before water is returned tothe environment.

We also find that the majority of future energy-related water impacts under a 2 °C climate policy areanticipated to occur in the electricity sector of devel-oping regions as a result of increasing electricitydemand. Widespread deployment of nuclear genera-tion and CCS technology to meet these requirementscould result in a water-intensive low-carbon energysystem that may contribute to localized water stress,

exacerbate conflict among competing water users, andthreaten the health of aquatic ecosystems throughthermal pollution. Particularly important therefore isthe development of integrated mitigation strategiesthat reduce both the carbon and water intensity ofelectricity generation in developing regions prone tosurface and groundwater stress that will be exacer-bated by increasing intersectoral competition and cli-mate change (vanVliet et al 2013).

Our analysis suggests that strategies combiningend-use energy efficiency (i.e., reducing energydemand)with a rapid scale-up of renewable electricitygeneration such as solar PV and wind can providemultiple co-benefits in terms of climate stabilization,reduced water demand and improved water quality.Previous work demonstrates an even broader range ofsustainability objectives covered by similar energytransformation pathways, including increased energysecurity and reduced human health impacts from airpollution (McCollum et al 2013). A key differencebetween these objectives is thatmost 2 °C climate poli-cies achieve improvements to energy security and airpollution, whereas many of the mitigation scenariosinvestigated here exacerbate water impacts.

Transitioning towards increased use of air-cooledtechnology for thermoelectric generation is found toreduce future water impact risks significantly. All 2 °Cscenarios investigated in this paper achieve lower end-of-century withdrawals than current conditions whenonce-through freshwater cooling technologies arereplaced with a combination of air and seawater cool-ing systems.We find relatively minimal impacts of thiscooling technology transition on electricity supplyefficiency. Yet, concerns surrounding the safety of air-cooled nuclear generation and incompatibilities withCCS technology could pose challenges to implementa-tion (Zhai et al 2011, Webster et al 2013). Air coolingtechnologies are also more expensive, and increase thecost of mitigating emissions (Webster et al 2013).Restricting expansion of nuclear and CCS technolo-gies based on anticipated water impacts is, however,risky and potentially costly in terms of climate stabili-zation (Rogelj et al 2013b).

Identifying locations that contain sufficient waterresources to support development and understandingfeedbacks between technology siting, water avail-ability, and deployment costs will require an enhancedspatial representation of the energy sector in inte-grated assessment models. This type of integratedapproach will prove critical when developing water,energy, and climate policy, especially in regions whererapid growth in both energy demand and waterdemands fromother sectors are anticipated.

Acknowledgments

The research has been supported by the EuropeanUnion’s Horizon 2020 Research and Innovation

9

Environ. Res. Lett. 11 (2016) 034011

Programme under grant agreement No 642147 (CD-LINKS) as well as by the European Union SeventhFramework Programme FP7/2007–2013 under grantagreement n° 308329 (ADVANCE). Simon Parkinsonwas also supported by a post-graduate scholarshipfrom the Natural Sciences and Engineering ResearchCouncil of Canada.

References

Bouckaert S, AssoumouE, Selosse S andMaïziN 2014Aprospectiveanalysis of waste heatmanagement at power plants andwaterconservation issues using a global TIMESmodel.Energy 6880–91

Byers EA,Hall JW andAmezaga JM2014 Electricity generationand coolingwater use: UKpathways to 2050Glob. Environ.Change 25 16–30

CameronC, YelvertonW,AkaR andWest J J 2014 Strategicresponses toCO2 emission reduction targets drive shift inUSelectric sector water useEnergy Strategy Rev. 4 16–27

ChuangY L, YangHHand LinH J 2009 Effects of a thermaldischarge from a nuclear power plant on phytoplankton andperiphyton in subtropical coastal waters J. Sea Res. 61197–205

Davies EG, Kyle P and Edmonds J A 2013An integrated assessmentof global and regional water demands for electricitygeneration to 2095Adv.Water Resour. 52 296–313

DelgadoA andHerzogH J 2012 Simplemodel to help understandwater use at power plantsWorking Paper, Energy InitiativeMassachusetts Institute of Technology

Döll P,Hoffmann-DobrevH, Portmann FT, Siebert S, Eicker A,RodellM and Scanlon BR 2012 Impact of water withdrawalsfrom groundwater and surfacewater on continental waterstorage variations J. Geodyn. 59 143–56

FlörkeM,Kynast E, Baerlund I, Eisner S,Wimmer F andAlcamo J2013Domestic and industrial water uses of the past 60 yearsas amirror of socio-economic development: a globalsimulation studyGlob. Environ. Change 23 144–56

Gerbens-LeenesW,Hoekstra AY and van derMeer TH2009Thewater footprint of bioenergy Proc. Natl Acad. Sci. USA 10610219–23

IEA 2012Water for energy: is energy becoming a thirstier resourceWorld EnergyOutlook 2012 (Paris: International EnergyAgency) ch 17

Kriegler E et al 2014The role of technology for achieving climatepolicy objectives: overview of the EMF 27 study on globaltechnology and climate policy strategiesClim. Change 123353–67

Kyle P,Davies EG,Dooley J J, Smith S J, Clarke L E,Edmonds J A andHejaziM2013 Influence of climate changemitigation technology on global demands of water forelectricity generation Int. J. Greenhouse Gas Control 13112–23

Macknick J, NewmarkR,HeathG andHallett KC 2012aOperational water consumption andwithdrawal factors forelectricity generating technologies: a review of existingliteratureEnviron. Res. Lett. 7 045802

Macknick J, Sattler S, Averyt K, Clemmer S andRogers J 2012bThewater implications of generating electricity: water use across

theUnited States based on different electricity pathwaysthrough 2050Environ. Res. Lett. 7 045803

McCollumDL, KreyV, Riahi K, Kolp P, Gruebler A,MakowskiM andNakicenovicN 2013Climate policies canhelp resolve energy security and air pollution challengesClim.Change 119 479–94

Mielke E, Anadon LD andNarayanamurti V 2010WaterConsumption of Energy Resource Extraction, Processing, andConversionBelfer Center for Science and International Affairs

Riahi K,Gruebler A andNakicenovicN 2007 Scenarios of long-termsocio-economic and environmental development underclimate stabilizationTechnol. Forecast. Soc. Change 74887–935

Riahi K et al 2012 Energy pathways for sustainable developmentGlobal Energy Assessment (Cambridge: CambridgeUniversityPress) ch 17

Rogelj J,McCollumDL,O’Neill BC andRiahi K 2013b 2020emissions levels required to limit warming to below 2 degreesCNat. Clim. Change 3 405–12

Rogelj J,McCollumDL, Reisinger A,MeinshausenMandRiahi K2013a Probabilistic cost estimates for climate changemitigationNature 493 79–83

Stewart R J,WollheimWM,Miara A, Vorosmarty C J, Fekete B,Lammers RB andRosenzweig B 2013Horizontal coolingtowers: riverine ecosystem services and the fate ofthermoelectric heat in the contemporary northeast USEnviron. Res. Lett. 8 025010

StrubeggerM,Totschnig G andZhuB 2004MESSAGE: a technicalmodel descriptionAchieving a Sustainable Global EnergySystem: Identifying Possibilities Using Long-term EnergyScenarios pp 168–213

SunN, Yearsley J, VoisinN and LettenmaierDP 2015A spatiallydistributedmodel for the assessment of land use impacts onstream temperature in small urbanwatershedsHydrol.Process. 29 2331–45

Tidwell VC,Macknick J, ZemlickK, Sanchez J andWoldeyesus T2014Transitioning to zero freshwater withdrawal in theUSfor thermoelectric generationAppl. Energy 131 508–16

Turchi C S,WagnerM J andKutscher CF 2010Water use inparabolic 475 trough power plants: summary results fromworleyparsons’ analysesTechnical ReportNational RenewableEnergy Laboratory

vanVlietMT, FranssenWH,Yearsley J R, Ludwig F,Haddeland I,LettenmaierDP andKabat P 2013Global river discharge andwater temperature under climate changeGlob. Environ.Change 23 450–64

vanVuurenDP, vanVliet J and Stehfest E 2009 Future bio-energypotential under various natural constraintsEnergy Policy 374220–30

WebsterM,Donohoo P andPalmintier B 2013Water-CO2 trade-offs in electricity generation planningNat. Clim. Change 31029–32

Wigley TMandRaper SC 2001 Interpretation of high projectionsfor global-meanwarming Science 293 451–4

Wimmer F, Audsley E,MalsyM, Savin C,Dunford R,Harrison PA,SchaldachR and FloerkeM2015Modelling the effects ofcross-sectoral water allocation schemes in EuropeClim.Change 128 229–44

ZhaiH, Rubin E S andVersteeg P L 2011Water use at pulverizedcoal power plants with post-combustion carbon capture andstorageEnviron. Sci. Technol. 45 2479–85

10

Environ. Res. Lett. 11 (2016) 034011