RIGHT:

URL:

CITATION:

AUTHOR(S):

ISSUE DATE:

TITLE:

Early retirement of power plants inclimate mitigation scenarios

Fofrich, Robert; Tong, Dan; Calvin, Katherine; De Boer,Harmen Sytze; Emmerling, Johannes; Fricko, Oliver;Fujimori, Shinichiro; Luderer, Gunnar; Rogelj, Joeri; Davis,Steven J

Fofrich, Robert ...[et al]. Early retirement of power plants in climatemitigation scenarios. Environmental Research Letters 2020, 15(9):094064.

2020-09-09

http://hdl.handle.net/2433/255250

© 2020 The Author(s). Published by IOP Publishing Ltd.; Original content from this workmay be used under the terms of the Creative Commons Attribution 4.0 license. Any furtherdistribution of this work must maintain attribution to the author(s) and the title of thework, journal citation and DOI.

Environmental Research Letters

LETTER • OPEN ACCESS

Early retirement of power plants in climate mitigation scenariosTo cite this article: Robert Fofrich et al 2020 Environ. Res. Lett. 15 094064

 

View the article online for updates and enhancements.

This content was downloaded from IP address 130.54.130.226 on 01/10/2020 at 10:43

A Self-archived copy inKyoto University Research Information Repository

https://repository.kulib.kyoto-u.ac.jp

Environ. Res. Lett. 15 (2020) 094064 https://doi.org/10.1088/1748-9326/ab96d3

Environmental Research Letters

OPEN ACCESS

RECEIVED

22 January 2020

REVISED

5 May 2020

ACCEPTED FOR PUBLICATION

27 May 2020

PUBLISHED

27 August 2020

Original content fromthis work may be usedunder the terms of theCreative CommonsAttribution 4.0 licence.

Any further distributionof this work mustmaintain attribution tothe author(s) and the titleof the work, journalcitation and DOI.

LETTER

Early retirement of power plants in climate mitigation scenariosRobert Fofrich1, Dan Tong1, Katherine Calvin2, Harmen Sytze De Boer3,4, Johannes Emmerling5,Oliver Fricko6, Shinichiro Fujimori6,7,8, Gunnar Luderer9,10, Joeri Rogelj6,11 and Steven J Davis1,121 Department of Earth System Science, University of California at Irvine, Irvine, CA, United States of America2 Joint Global Change Research Institute, Pacific Northwest National Laboratory, College Park, MD, United States of America3 PBL Netherlands Environmental Assessment Agency, 2594 AV The Hague, The Netherlands4 Copernicus Institute for Sustainable Development, Utrecht University, Utrecht, The Netherlands5 RFF-CMCC European Institute on Economics and the Environment (EIEE), Centro Euro-Mediterraneo sui Cambiamenti Climatici,Via Bergognone 34, I-20144 Milan, Italy

6 Energy Program, International Institute for Applied Systems Analysis (IIASA), 2361, Laxenburg, Austria7 Center for Social and Environmental Systems Research, National Institute for Environmental Studies (NIES), Tsukuba, Japan8 Department of Environmental Engineering, Graduate School of Engineering, Kyoto University, Nishikyo-ku, Japan9 Potsdam Institute for Climate Impact Research (PIK), Member of the Leibniz Association, Potsdam, Germany10 Chair of Global Energy Systems, Technische Universitat Berlin, Straße des 17. Juni 135, 10623 Berlin, Germany11 Grantham Institute for Climate Change and the Environment, Imperial College London, London, United Kingdom12 Department of Civil and Environmental Engineering, University of California at Irvine, Irvine, CA, United States of America

E-mail: rfofrich@uci.edu

Keywords: climate change, energy systems, power plants, representative concentration pathways, integrated assessment model, sharedsocioeconomic pathways, climate policy

Supplementary material for this article is available online

AbstractInternational efforts to avoid dangerous climate change aim for large and rapid reductions of fossilfuel CO2 emissions worldwide, including nearly complete decarbonization of the electric powersector. However, achieving such rapid reductions may depend on early retirement of coal- andnatural gas-fired power plants. Here, we analyze future fossil fuel electricity demand in 171energy-emissions scenarios from Integrated Assessment Models (IAMs), evaluating the implicitretirements and/or reduced operation of generating infrastructure. Although IAMs calculateretirements endogenously, the structure and methods of each model differ; we use a standardapproach to infer retirements in outputs from all six major IAMs and—unlike the IAMsthemselves—we begin with the age distribution and region-specific operating capacities of theexisting power fleet. We find that coal-fired power plants in scenarios consistent with internationalclimate targets (i.e. keeping global warming well-below 2 ◦C or 1.5 ◦C) retire one to three decadesearlier than historically has been the case. If plants are built to meet projected fossil electricitydemand and instead allowed to operate at the level and over the lifetimes they have historically, theroughly 200 Gt CO2 of additional emissions this century would be incompatible with keepingglobal warming well-below 2 ◦C. Thus, ambitious climate mitigation scenarios entail drastic, andperhaps un-appreciated, changes in the operating and/or retirement schedules of powerinfrastructure.

Introduction

Among scenarios that succeed in stabilizing globalmean temperatures at less than 2 ◦C warmer thanthe preindustrial era, CO2 emissions from the powersector decrease rapidly in the coming decades, inalmost all cases reaching net-zero before mid-century[1–5]. Such rapid and complete decarbonization

entails similarly rapid turnover of historically long-lived electricity-generating infrastructure. Coal- andgas-fired power plants have historically operated for39 and 36 years (s.d.14 and 13 years), respect-ively [6]. However, in Integrated Assessment Mod-els (IAMs), the decision of when to retire a gener-ator is primarily economic, e.g. based on marginaloperating costs, revenues, and the levelized costs of

© 2020 The Author(s). Published by IOP Publishing Ltd

A Self-archived copy inKyoto University Research Information Repository

https://repository.kulib.kyoto-u.ac.jp

Environ. Res. Lett. 15 (2020) 094064 R Fofrich et al

Figure 1. Schematic of modelling approach. Figure shows a hypothetical scenario to illustrate our methodological approach and isnot representative of any specific integrated assessment model or shared socioeconomic pathway. Here we see, given a futureelectricity demand from coal- and gas-fired power plants in an integrated assessment model scenario (black curves), it may benecessary to build additional generating capacity (colored squares), whose operation may eventually exceed demand withcorresponding ‘overshoot’ of emissions (hatched squares). Nonetheless, this schematic represents the model in its simplest formand does not capture the full extent of model ensembles.

new generating infrastructure [7–9]. IAM mitiga-tion scenarios reconcile these economics with swiftdecarbonization of the electricity sector by modelingboth policy-driven increases in the operational costsof CO2-emitting power plants and rapidly decreasingcosts of non-emitting sources of electricity [10, 11].In reality, lawmakers may follow a similar approach,incentivizing the early closure of plants or severelyreducing their operating hours by imposing strictregulations that increase their operating costs relat-ive to non-emitting competitors. Examples of spe-cific policies include setting a price on carbon, disal-lowing major maintenance (e.g. New Source Reviewin the United States), or subsidizing non-emittingtechnologies (e.g. renewable production tax credits).However, economics are not the sole determinantof power plant retirements, as there are numerousexamples of fossil power plants nowoperating at a loss[12–14]. This suggests that more direct regulationssuch as an outright ban of a given fossil technology ormandating the early closure of certain power plantsmay be necessary. Nonetheless, given the initial cap-ital costs of fossil fuel electricity generating capacityare typically $200–5000 per kW and installed fossilcapacity worldwide is today ∼4000 GW [9, 15, 16],the premature retirement of power generating infra-structure could result in the loss of trillions of dol-lars of capital investment and future returns, andperhaps even jeopardize the stability of financialsystems if not adequately managed and anticipated[17–20]. Moreover, losses from early retirement offossil electricity generating assets may ultimately beborne by the rate- and tax-paying populace. For thesereasons, the socioeconomic and political repercus-sions that arise from very early retirement of coal-and gas-fired power plants may be challenging toovercome.

Several previous studies have estimated the CO2

that will be emitted by existing and proposed energy

infrastructure if it is operated for historical averagelifetimes [6, 8, 16]. Others have used IAMs in vari-ous ways: using scenarios as a guide to future fossilcapacity [21], adding plant lifetime as an exogenousconstraint within a model [22], or evaluating theinfrastructural inertia of emissions in a designedmulti-model experiment [23]. However, prior workhas generally focused on differences in emissionsrelated to the lifetime, operation, or commission-ing of generating infrastructure. Here, we also takethe opposite perspective: what do the rapid emis-sions reductions in mitigation scenarios imply forthe lifetime, operation, and commissioning of gener-ating infrastructure? Specifically, how severely mustthe lifetime or operation of power plants be abbrevi-ated or curtailed, respectively, in order to achieve theemissions decreases (i.e. mitigation rates) in differentscenarios and regions? Although the answers to thesequestions can be explicitly calculated by some IAMs,modeling approaches between IAM vary, retirementsare endogenous to the models, and retirement ratesare not reported—or even tracked—by all modelinggroups.

Here, using detailed data of currently exist-ing power plants worldwide [24] in addition toelectricity and emissions outputs from six majorintegrated assessment models, we analyze coal-and natural gas-fired power plant utilization ratesand lifetimes as embedded in 171 recent scen-arios, spanning three levels of emissions mitiga-tion (1.9, 2.6, and 4.5 W m−2 of radiative forcing;i.e. trajectories likely to avoid 1.5 ◦C, 2 ◦C, and3 ◦C of mean warming this century), and five dif-ferent socioeconomic trajectories (SSPs) [25]. Weexplicitly exclude oil-fired power generators fromour analysis since they compose less than 5% ofglobal electricity generating capacity [26]. Furtherdetails of our analytic approach are in the Meth-ods and supplementary information (available online

2

A Self-archived copy inKyoto University Research Information Repository

https://repository.kulib.kyoto-u.ac.jp

Environ. Res. Lett. 15 (2020) 094064 R Fofrich et al

Figure 2. Inertia in power sector emissions. Future emissions from coal- and gas-fired power plants in the 1.9, 2.6, and4.5 W m−2 radiative forcing scenarios (black curves) often decrease more rapidly than emissions from power plants which atregion-specific mean capacity factors and power plant lifetimes ranging from 10 years to 60 years (colored curves). The thin linesshow each IAM-SSP combination, and the bold lines show the median value of all IAM-SSP projections. Given the age structureof now-existing energy infrastructure, ambitious mitigation pathways such as 1.9 and 2.6 W m−2 imply very short power plantlifetimes, particularly for coal-fired units.

at (stacks.iop.org/ERL/15/094064/mmedia). Figure 1summarizes howour analyseswere conducted schem-atically. In this figure we only show the simplestapproach to facilitate the readers understanding

of our methodology. Here we assume a uniformoperating lifetime (e.g. 40 years in figure 1(a)) andcapacity factor (e.g. 70% in figure 1(a)). In addi-tion, we evaluate whether and when fossil fuel- and

3

A Self-archived copy inKyoto University Research Information Repository

https://repository.kulib.kyoto-u.ac.jp

Environ. Res. Lett. 15 (2020) 094064 R Fofrich et al

Figure 3. Annual mean emission mitigation rates, cumulative emissions and emission overshoot in energy-emission scenarios.Cumulative CO2 emissions from coal- and gas-fired power plants in the 1.9, 2.6, and 4.5 Wm−2 radiative forcing trajectories overthe 21st century (a–c). Cumulative emissions increase as power plants lifetimes are prolonged and as climate mitigation goalswane. Annual emission reductions from coal and gas electrical generators decline with an increase in assumed power plantlifetime and with increased inertia from electricity production (d–f). Differences between SSP emissions projections andemissions under different lifetime assumptions (g–i). Dashed vertical line indicates the historical mean lifetime whereas the whitedashed line is the cumulative emission mean across all IAM-SSPs for each of the forcing scenarios. Color intensity indicates the50th—95th percentile cumulative emissions for all of the IAM-SSPs. The light horizontal line represents the median cumulativeemission overshoot value, if power plants follow historical mean lifetime trends. The cumulative emission overshoot underdifferent lifetime assumptions decreases as the radiative imbalance increases.

region-specific electricity demand in each IAM scen-ario (black curves) will require new capacity to becommissioned (colored squares) if existing capacity(gray squares) is not able to meet the projected fossilelectricity need. As fossil electricity demand declineswithin the IAMs in the future, we quantify the extentto which there would be excess generating capacitygiven the assumed lifetime and capacity factor ofoperating power plants (black-hatched squares). Byfurther assuming a carbon emissions factor (CO2

per unit electricity generated) in line with historicalestimates, we can in turn quantify the potential emis-sions associated with such excess capacity. Assumedlifetime, capacity factor, and carbon emission factorsare varied in repeated analyses (e.g. figures 1(b) and(c)). We analyze model projections using fixed life-times and capacity factors to project all plausible val-ues of future emissions. Additionally, we vary power

plant operating conditions in each subsequent annualtime step as a sensitivity test for our results. However,this added flexibility to the initial operational con-ditions of power generating infrastructure had verylittle impact on our overall results. For context, table 1compares operating conditions and constraints oninfrastructure retirements within each of the sixIAMs.

In figure 2, the black curves show the annual CO2

emissions from coal- and gas-fired electricity genera-tion, as projected by the integrated assessment mod-els, for all SSPs under different levels of future warm-ing used in this study (i.e. radiative forcing of 1.9,2.6, and 4.5 W m−2). In comparison, colored curvesshow our calculated emissions if power plant life-times are assumed to be 10, 20, 30, 40, 50, or 60 years(purple, blue, green, yellow, orange, and red, respect-ively). Here we also assume historical mean capacity

4

A Self-archived copy inKyoto University Research Information Repository

https://repository.kulib.kyoto-u.ac.jp

Environ. Res. Lett. 15 (2020) 094064 R Fofrich et al

Table 1. Integrated Assessment Model Assumptions. Regional averaged values for each of the integrated assessment models used withinthis study. However, as the IAMs continue to evolve so do the underlying parameters. Thus, values represented in this table may changeover time as newer versions of IAMs are released.

Lifetime (years) Capacityfactor (max-

imum/minimum)

Depreciationof capital

rate (averagepercent per year)

Carbon intensity(range acrosstechnologies,regions, years,and SSPs)

CoalAIM/CGE 35 60% 4% Different

across regionsGCAM 60 80%–85%

depending ontype of plant

643 to 1233 gCO2per kWh,

depending ontechnology,region, year

IMAGE 40 Dependingon relative

operational costs(∼85% till 0%)

Capacity getsretired after40± 5 yearsof operation

Different perregion, year,technology

MESSAGE-GLOBIOM

30 67%–85% 5% 724–1302 gCO2per kWh

REMIND-MAGPIE

40 75%–80% Non-linear Different perregion, year,technology;regional fleet

averages of 738–1140 g kWh−1

in 2015WITCH-GLOBIOM

40 85% 2.8% 699 to1390 gCO2 kWh−1,

depending ontechnology,region, year

GasAIM/CGE 30 70% 4% Different

across regionsGCAM 60 for existing

gas plants, 45for new plants

80%–85%depending ontype of plant

274 to 720 gCO2per kWh,

depending ontechnology,region, year

IMAGE 40 Dependingon relative

operational costs(∼90% till 0%)

Capacity getsretired after40± 5 yearsof operationor via early

retirement in caseof relatively highoperational costs

Different perregion, year,technology

MESSAGE-GLOBIOM

30 58%–85% 5% 260–850 gCO2 kWh−1

REMIND-MAGPIE

35 55%–65% Non-linear Different perregion, year,technology;regional fleet

averages of 328–547 g kWh−1

in 2015WITCH-GLOBIOM

25 70% 4.4% 354 to1000 gCO2 kWh−1,

depending ontechnology,region, year

5

A Self-archived copy inKyoto University Research Information Repository

https://repository.kulib.kyoto-u.ac.jp

Environ. Res. Lett. 15 (2020) 094064 R Fofrich et al

Figure 4. Excess CO2 emissions from coal and gas-fired power plants. Differences between mean IAM emissions projections andmean estimated CO2 emissions under different capacity factor and lifetime assumptions. The panel rows represent the threedifferent levels of radiative forcing (1.9, 2.6, and 4.5 W m−2) while the panel columns show the difference between coal- andgas-fired power plants. Color shading indicate a range of capacity factors ranging from 35%–75%. Dashed vertical line representsthe historical mean power generator lifetime of 37 years whereas the white dashed line moving along the x-axis represents thehistorical mean capacity factor.

and carbon emissions factors, see tables S1–6, how-ever we vary power plant operational conditions insubsequent calculations to test impacts on our res-ults. In all cases, bold curves represent the medianof all global integrated assessment model scenarios(n= 171).

We see the median IAM emissions (black curves)generally decrease more quickly than the emissions

we estimate if plants were to operate for more than30 years (green curves), especially in the case of coal-fired plants and under the more ambitious (lowerwarming) scenarios (figure 2). For example, figure2(a) shows that median emissions, assuming coal-fired generator lifetimes greater than 30 years, donot decline as rapidly as the median IAM projections(bold black curve) for the 1.9 W m−2 scenario. The

6

A Self-archived copy inKyoto University Research Information Repository

https://repository.kulib.kyoto-u.ac.jp

Environ. Res. Lett. 15 (2020) 094064 R Fofrich et al

Figure 5.Maximum power plant lifetime under different electricity-emission scenarios. Under ambitious climate changescenarios, fossil powered electricity generating infrastructure retire much earlier than they have historically. Here we present themaximum obtainable lifetime under different electricity demand scenarios for three levels of radiative forcing (radiative forcing1.9, 2.6, and 4.5 W m−2). Error bars show the full range of power retirements under different capacity factor assumptions.

differences between the black IAMcurves and our cal-culated curves reflects the magnitude of such excessemissions, which consistently increase as longer life-times are considered. However, the scenarios fromdifferent IAMs and SSPs can result in consider-ably different cumulative emissions, with greatermodel spread under higher warming scenarios (fromleft to right in figures 3(a)–(c)). For instance, inthe lower warming (i.e. likely to avoid 1.5 ◦Cand 2 ◦C) scenarios, cumulative emissions averagedacross models and assumed lifetimes are greatest forSSP2 (‘middle-of-the-road’; blue), followed by SSP5(‘fossil-fueled development’; pink) and least for SSP1(‘sustainability’; green) and SSP4 (‘inequality’; paleorange). See Methods or ref [27] for further dis-cussion on how the SSPs differ. Averaging acrossmodels, for a given lifetime, cumulative emissionsvary by 27%, 30%, and 36% across SSPs in the dif-ferent warming scenarios, respectively. In compar-ison, the average variation in cumulative emissionsamong models for a given SSP and lifetime are 31%,45%, and 48% in the different warming scenarios,respectively.

The longer the assumed lifetime of power plants,the lower mean mitigation rates (defined here asthe annual percent reduction in CO2 emissions from2017–2050) will be, figures 3(d)–(f). Since meanmit-igation rates are inversely related to future warming,this relationship illustrates the temporal constraintsimposed by infrastructural inertia. For example, inthe scenarios likely to bring back warming to below1.5 ◦C by 2100 (SSPx-1.9 scenarios from ref. [11]),integrated assessment model outputs average 6% peryear reductions in emissions from coal- and gas-firedpower plants (dotted gray line), but mean mitigation

rates when assuming plant lifetimes of 30 or moreyears decrease to <3% per year (figure 3(d)). Simil-arly, model outputs average 3.7% per year reductionsin scenarios likely to avoid 2 ◦C (SSPx-2.6, dotted grayline), but meanmitigation rates when assuming plantlifetimes of 30 or more years decrease to <2% peryear (figure 3(e)). Thus, allowing fossil-fired powerinfrastructure to operate for more than 30 years frominitial commissioning is incompatible with the rapidmitigation rates achieved in the IAMs.

Since climate change is proportional to society’scumulative emissions, we were interested in quanti-fying the amount of emissions over the IAMs (hereby‘cumulative overshoot’) when power generators areoperated for different periods of time. We find thecumulative overshoot increase along with assumedlifetimes but are also substantially greater in the lowerwarming scenarios (figures 3(g)–(i)). For instance,if we assume power generators will follow historicaloperating norms, a lifetime of 37 years and meancapacity factor (dashed lines), the cumulative over-shoot rises from a median 112 Gt CO2 in 4.5 W m−2

scenarios, to 188 Gt CO2 in 2.6 W m−2 scenarios,to 220 Gt CO2 in 1.9 W m−2 scenarios. Given thattotal cumulative emissions averages just 182.5 Gt CO2

in 1.9 W m−2 scenarios, an additional 220 Gt CO2

represents an overshoot of 220.5% and is roughlyequivalent to the entire fossil electricity CO2 budgetin the 2.6 W m−2 scenario. We find the similaritybetween the 1.9 and 2.6 W m−2 scenarios (figure 4)largely result from the age distribution of the exist-ing power fleet. In both cases, the IAM scenarios res-ult in immediate reductions to global CO2 emissionsbut do not consider the power infrastructure life-times of operating plants. Using our methods, but

7

A Self-archived copy inKyoto University Research Information Repository

https://repository.kulib.kyoto-u.ac.jp

Environ. Res. Lett. 15 (2020) 094064 R Fofrich et al

following the 2.6 W m−2 scenario requires modestdeployment of new fossil capacity resulting in a sim-ilar overshoot. Nonetheless, these findings indicatethe extent to which the low cumulative emissions inambitious mitigation scenarios are the result of earlyretirement of coal- and gas-fired power plants. Inaddition, the similarity of the IAM electricity path-ways while achieving different levels of radiative for-cing indicate that a substantial reduction of annualCO2 emissions from other industries is required toreach the 1.9 Wm−2 pathway. Figure 5 highlights thefull range of power plant retirements under a rangeof capacity factor assumptions. If coal and gas powerplant operations are severely curtailed, and climatemitigation targets are relaxed, then these plants mayoperate at similar lifetimes as they have historically.In contrast, these plants would have to retire decadesearlier than they have in the past if we are to meetmore ambitious climate warming trajectories.

In turn, supplementary figure 1 shows how keyregions contribute to the cumulative overshoot inlower warming scenarios (averaging across the val-ues for 1.9 and 2.6 W m−2 shown in figures 3(g)and (h)). In comparison to the other regions shown,overshoots increase most dramatically in China whenlonger lifetimes of power plants are assumed. Thisis consistent with previous unit-level inventories ofemissions which have shown that half of now-existingcoal-fired generating capacity is in China, and mostly<15 years old [28]. Supplementary figure S1(a)reveals the extent to which model scenarios anticip-ate the retirement of these Chinese plants before theyreach 20 years of age. Similarly, early retirementsare required to avoid substantial overshoots in otherregions, but the magnitude of overshoot when an his-torical lifetime of 37 years is assumed are roughly53%, 26% and 87% less in India, theU.S. andWesternEurope than in China, respectively.

Supplementary figure 2 acts as sensitivity testto our projected emissions from allowing additionalflexibility in initial power plant operational condi-tions. For example, varying assumptions of plantlifetime and capacity factor by 25% has a similareffect on estimated cumulative emissions, regardlessof radiative forcing or SSP (Supplementary figure S2).However, both lifetime and capacity factor becomeless important in higher warming scenarios, and theassumed carbon intensity of electricity becomes adominant factor (Supplementary figure S2).

1. Discussion and conclusions

Our results suggest that climate scenarios which arestabilize global temperatures in the range of 1.5 ◦Cto 2 ◦C or below, retire coal- and gas-fired plantsdecades before their technical or historical lifetimeshave been reached. Although it is generally under-stood that CO2 emitting infrastructure will need to beswiftly decommissioned in order to mitigate the most

extreme consequences of climate change, the extent towhich climatemitigation scenarios rely on the prema-ture retirement of existing plants and the curtailmentof future construction is not widely known. SinceIAMs conduct power plant retirements endogen-ously, the rates and processes that dictate these retire-ments seem obscure to many who wish to interpretIAM results [29]. In addition, the IAM projectionstypically begin in 2005 and without incorporatinginformation about the current installed fossil capacityor age distribution of fossil fuel-fired plants. Thus,climate mitigation scenarios may underestimate theinertia of emitting infrastructure. As a result of theIAM structure, the operating power capacity and pro-jected mitigation rates in their scenarios can quicklydiverge from the realities of the existing fossil fleet andcan vary greatly between IAMs and SSPs.

The mitigation rates observed within IAMs areunprecedented and thus represent a potential chal-lenge to society, particularly with the continueddeployment of coal-fired power plants around theglobe [30]. If coal-fired power generators are notretired early (or their capacity factors drasticallyreduced), then mitigation rates will fall behind IAMscenarios (figures 3(d) and (e)) and cumulative emis-sions will rise sharply (figures 3(a), (b), (g), and(h)), thus undermining the ability to achieve lower-warming targets without additional compensatorydecreases in emissions from other sources [26, 27].Although negative emissions are represented withinthe integrated assessment models, our results high-light that longer power plant lifetimes would requirean even larger negative emissions than the prodigiousquantities already present in some of the more ambi-tious mitigation scenarios (which are in some casesmany Gt CO2 per year) [31]. Moreover, the need forshortened infrastructure lifetimes is particularly crit-ical in China, where coal-fired generating capacity isboth young and large [16].

Given the established relationship of cumulat-ive carbon budgets and climate warming [32–35],prior studies have estimated and compared ‘commit-ted’ emissions over the expected lifetime of emit-ting infrastructure [6, 8, 16, 36]. Many climate mit-igation scenarios thus optimize operating and retire-ment schedules of fossil-fueled infrastructure to lowertheir cumulative carbon emissions (hence attain-ing lower carbon budgets and establishing lowerwarming trajectories) by prioritizing economic con-ditions where costs of the power sector are equalto revenues from electrical generation rather thanreflecting the inertia of the power fleet which isalready in existence today. In actuality, decommis-sioning trillions of dollars’ worth of privately-ownedcapital after only 25% of its anticipated life haselapsed will present enormous political and eco-nomic challenges. Indeed, it is these challenges, col-lectively, that represent the infrastructural inertia(i.e. carbon lock-in) [9, 16, 36].

8

A Self-archived copy inKyoto University Research Information Repository

https://repository.kulib.kyoto-u.ac.jp

Environ. Res. Lett. 15 (2020) 094064 R Fofrich et al

While the IAMs serve as a powerful tool, allow-ing users to gain insight regarding a particular sec-tor, the mechanisms behind endogenous calculationsare often seen as black boxes by the broader scientificcommunity leading some to question their methodsas inscrutable [29]. Thus, by using a standardizedmethod to quantify the implicit lifetimes of powerplants within these climate mitigation scenarios, ouranalysis provides a transparent process while demon-strating the extent to which lower warming scen-arios may be contingent upon the early retirement ofpower sector infrastructure. In many cases, deliber-ately planned retirement of coal- and gas-fired powerplants are necessary in mitigation scenarios whichproject limited growth in demand for fossil-fuel elec-tricity. If instead, the deployment of fossil fuel powercapacity is continued in the upcoming years, stabil-izing global mean temperatures at less than 2 ◦C rel-ative to the preindustrial will require even shorterretirement ages than those achieved within climatemitigation scenarios. Nonetheless, our results sug-gest that these targets can only be achieved througha strategic manipulation of installed coal- and gas-fired power capacity, generator lifetimes, and capa-city factors (e.g. retiring certain plants prematurelyor severely curtailing their usage while extending thelifetime of others until renewable electricity gener-ating technology is deployed locally at scale). Thus,if current power sector trends continue, this maynecessitate economically costly options—e.g. strand-ing fossil electrical assets, retrofitting existing plantswith CCS, or offsetting increased emissions throughmass deployment of carbon dioxide removal techno-logies [5, 37], which ultimately may come at a higherexpense than early retirement.While the value of suchgenerating capital and the total cost to society are rep-resented and depreciated within these scenarios, thedistribution of these costs is not. Therefore, lost rev-enues and profitability for plant owners and local gov-ernments, or job losses for workers might prove pro-hibitively high.

It should be noted that some of our projections offuture emissions reported here do not allow lifetimesand capacity factors to vary over time, across regions,or between different generating assets which is in con-trast to the flexibility allowed in power plant oper-ational conditions both in the integrated assessmentmodels and the real world. Thus, insofar as capacityfactors and lifetimes may in reality decrease over thelifetime, operation, and retirements may be strategic-ally scheduled, and plants might be mothballed andre-operated. Thus, the overshoot we project shouldbe interpreted to reflect the capacity-weighted averagelifetime and may be overestimated. However, we findit crucial to demonstrate the incapability of continuedinvestments in fossil fuel power infrastructure withmore ambitious climate mitigation scenarios ratherthan focus on any one single lifetime trajectory. Thatis, because it is newly commissioned power plants

that create the greatest inertia and scenario overshoot.While in some cases inertia and emissions could beavoided by extending the life of existing and due-to-retire plants, such that new plants will not have to bebuilt (and the older plants can be more readily retiredto rapidly decrease emissions), achieving such flexib-ility in reality would depend upon clear foresight ofboth regional electricity demand and global climate-energy policies, as well as rational economic beha-vior on the part of utilities and power plant own-ers whom historically have not been transparent intheir decisions [38, 39]. Nonetheless, decarbonizingthe global power sector is currently technically andeconomically feasible given proven technology but iscontingent on the increased investment and construc-tion of low-carbon technology and infrastructure aswell as passing legislation regulating carbon emit-ting technologies [40]. While costly, the co-benefitsto society often outweigh the overall financial bur-dens that result from a swift retirement of pollutingplants [41]. Thus, policy makers should immediatelybegin to phase out fossil-fired power plants by sup-porting low-carbon energy infrastructure while sim-ultaneously implementing legislation that’s unfavor-able for continued fossil fuel use. However, in reality,governments have been observed taking the oppos-ite approach, choosing instead to prop up economic-ally unstable power plants through subsidies and/orby passing industry favorable regulations in order tominimize the socioeconomic consequences of plantclosures and ultimately prolonging the infrastructuralinertia of these plants [39].

Thus, in conclusion, power sector capital thatis amassed over decades will also take decades toretire unless its value is sacrificed, and lower-warmingscenarios often demand such sacrifice. Which policymechanisms force early retirements may ultimatelydeterminewhowill bear the economic losses. In juris-dictions with strict climate policies, proactively limit-ing the time period that new coal- and gas-fired plantswill be allowed to operate might forestall investmentsthat would otherwise either contribute to emissionsovershoot or else be forced to retire early at greatexpense. In the future, operating lifetimes and eco-nomic implications of CO2 emitting-infrastructureshould be consideredwhen formulating future energyinvestments that are consistent with existing climatepolicies so that investors may determine the compat-ibility of their planned energy infrastructure invest-ments with different scenarios of climate change andfully understand the risks of their monetary invest-ments [18, 38].

2. Methods

2.1. Existing and historical infrastructureWe use the Global Power Plant Emissions Database(GPED) to analyze historical coal and gas powerplants that are currently operating. We quantify

9

A Self-archived copy inKyoto University Research Information Repository

https://repository.kulib.kyoto-u.ac.jp

Environ. Res. Lett. 15 (2020) 094064 R Fofrich et al

the annual electrical generation, installed nameplatecapacity, yearly averaged emission intensities, andannual mean capacity factor of all existing and pastpower plants. For currently operating generators, weidentify current installed capacity in each region andthe year each was commissioned, and project theexpected year of retirement based on an assumed life-time.

2.2. Power infrastructure commissioned in futureRegional scenarios of future electricity projectionswere produced for each of the Shared SocioeconomicPathways (SSPs) by the Asia-pacific Integ-rated Model/Computable General Equilibrium(AIM/CGE), Global Change Assessment Model(GCAM), Integrated Model to Assess the GlobalEnvironment (IMAGE), the Model of Energy Sup-ply Strategy Alternatives and their General Envir-onmental impacts—Global Biosphere Management(MESSAGE-GLOBIOM), Regional Model of Invest-ments and Development—Model of AgriculturalProduction and its Impact on the Environment(REMIND-MAGPIE), and World Induced Tech-nical Change Hybrid—Global Biosphere Manage-ment (WITCH-GLOBIOM) integrated assessmentmodels (IAMs). Each IAM uses different number ofregions to represent global society and classifies theseregions based on their socioeconomics, geopolitics,and stage in economic development of the nationsrepresented. A full list of IAM regions and associatedhistorical mean capacity factors and carbon intens-ities is provided within the supplementary informa-tion, tables 2–7. We quantify existing power generat-ing infrastructure, electricity demand, and generatoroperating conditions using the same regional classi-fications as represented in each IAM.We then projectthe need for new electricity generating capacity byestimating the difference between IAM projectionsand existing electrical capacity in each world regionand SSP-model-radiative forcing trajectories.

Repeated analyses vary the assumed lifetimes ofcoal- and gas-fired power plants 10–60 years andcapacity factors from 35%–75%, applicable to bothexisting generators and any infrastructure commis-sioned in the future. In our standardized approach,power generators are phased out once their expectedoperational lifetime has elapsed. New power gener-ators are only built if the annual power supply dipsbelow annual power demand, which can occur whenexisting power infrastructure is retired or if there isa sustained increase in power demand projected bythe IAMs. Newly constructed generators are assumedto have the same operating conditions as the corres-pondingmodel run. Nonetheless, we calculate the 1.9and 2.6 W m−2 radiative forcing scenarios requiredvery little deployment of new coal-fired power plants,instead most of the overshoot observed in our resultscome from existing power infrastructure with theexemptions of a few regions globally.

2.3. EmissionsWe convert our estimates of electricity generationto carbon dioxide emissions using IAM electricityprojections, our energy calculations under differ-ent lifetime assumptions, and IAM regional meanhistorical carbon intensities ranging from 387–1381.4 gCO2 kWh−1. Here we analyze 18 810 ofindividual IAM regional coal and gas electricity scen-arios and categorically applied the correspondingcarbon intensity. A detailed list of IAM regional meancarbon intensities can be found in the supplementaryinformation, tables 2–7. Additionally, we use a linearregression approach and looked at the annual emis-sion reductions 2017 to 2050, to determine the annualemission mitigation rates of each IAM-SSP includedin this study. For each radiative forcing pathway,cumulative emissions overshoot was determined bytaking the difference between the cumulative emis-sion projection and the cumulative emissions traject-ories under the various power plant lifetime assump-tions used for this study. In each RF, cumulativeemissions are calculated by model, SSP, and lifetimeassumption individually then separated by their stat-istical distribution thus identifying the probability ofthe emissions trajectory.

2.4. Regional analysisWeanalyze regional emissions under each of the IAMsincluded in this study using the mean IAM regionalcapacity factors and carbon emissions intensities. Ineach case, we calculate the cumulative emission over-shoot for both coal-fired and natural gas electricitygeneration individually by RF, IAM, and SSP. Weseparate the cumulative emission overshoot by theirstatistical distribution to quantify the likelihood ofthis emission projection and plot the median cumu-lative carbon dioxide emissions in each case. Addi-tionally, we identify the magnitude of CO2 emis-sion overshoot for each region based on historicalmedian power plant lifetimes of 37 years. Regionalcalculations are based on IAM regional classifica-tions and are aggregated to quantify global energyand emissions. In each case, we analyze global emis-sions overshoot for each of the radiative forcing tra-jectories included in this study. Here we calculatedthe overshoot and again vary the historical capa-city factors by 35%–75% and vary the power plantlifetimes from 10–60 years. Using the GPED data-base, we estimate the historical capacity factors tobe ∼65% and ∼55% for coal and gas power plants,respectively.

Acknowledgments

R.F. and S.J.D. were supported by the U.S. NationalScience Foundation (INFEWS grant EAR 1639318).D.T. was supported by NASA’s IDS program(80NSSC17K0416).

10

A Self-archived copy inKyoto University Research Information Repository

https://repository.kulib.kyoto-u.ac.jp

Environ. Res. Lett. 15 (2020) 094064 R Fofrich et al

Contributions

RF and SJD designed the study; RF led the analysiswith additional input and data support from SJD, DT,KC, HSB, JE, OF, SF, GL, and JR; RF led the writingwith input and revisions from all coauthors.

Data availability

The data that support the findings of this study areavailable from the corresponding author upon reas-onable request

Conflict of interest

The authors declare no competing interests.

ORCID iDs

Robert Fofrich https://orcid.org/0000-0001-6418-3692Katherine Calvin https://orcid.org/0000-0003-2191-4189Johannes Emmerling https://orcid.org/0000-0003-0916-9913Shinichiro Fujimori https://orcid.org/0000-0001-7897-1796Gunnar Luderer https://orcid.org/0000-0002-9057-6155Joeri Rogelj https://orcid.org/0000-0003-2056-9061Steven J Davis https://orcid.org/0000-0002-9338-0844

References

[1] Davis S J et al 2018 Net-zero emissions energy systemsScience 360 1419

[2] Rogelj J, Schaeffer M, Meinshausen M, Knutti R, Alcamo J,Riahi K and Hare W 2015 Zero emission targets as long-termglobal goals for climate protection Environ. Res. Lett.10 105007

[3] Williams J H, DeBenedictis A, Ghanadan R, Mahone A,Moore J, Morrow W R, Price S and Torn M S 2012 Thetechnology path to deep greenhouse gas emissions cuts by2050: the pivitol role of electricity Science335 53–59

[4] Audoly R, Vogt-Schilb A, Guivarch C and Pfeiffer A 2018Pathways toward zero-carbon electricity required for climatestabilization Appl. Energy 225 884–901

[5] Luderer G et al 2018 Residual fossil CO2 emissions in1.5–2 ◦C pathways Nat. Clim. Change 8 626–33

[6] Davis S J, Caldeira K and Matthews H D 2010 Future CO2

emissions and climate change from existing energyinfrastructure Science 329 1330–3

[7] Taylor M J and Fuller L C 1986 Coal-fired Electric PowerPlant Life Extension: An Overview (Oak Ridge, TN: OakRidge National Laboratory)

[8] Davis S J and Socolow R H 2014 Commitment accounting ofCO2 emissions Environ. Res. Lett. 9 084018

[9] Seto K C, Davis S J, Mitchell R B, Stokes E C, Unruh G andÜrge-Vorsatz D 2016 Carbon lock-in: types, causes, andpolicy implications Annu. Rev. Environ. Resour.41 425–52

[10] Rogelj J, Luderer G, Pietzcker R C, Kriegler E, Schaeffer M,Krey V and Riahi K 2015 Energy system transformations forlimiting end-of-century warming to below 1.5 ◦C Nat. Clim.Change 5 519–27

[11] Rogelj J et al 2018 Scenarios towards limiting global meantemperature increase below 1.5 ◦C Nat. Clim. Change8 325–32

[12] Wamsted D, Seth F and Schlissel D Coal Outlook 2019 -Domestic Market Decline Continues, The Institute forEnergy Economics and Financial Analysis (IEEFA)

[13] Gray M, Sebastian L, Watson L and Kok I 2018 Poweringdown coal - Navigating the economic and financial risks inthe last years of coal power, Carbon Tracker Initiative

[14] Gimon Eric O B M, Christopher C and Sarah M 2019 TheCoal Cost Crossover: Economic Viability of Existing CoalCompared to New and Local Wind and Solar Resources,Energy Innovation

[15] Baron R and Fischer D 2015 Divestment and stranded assetsin the low-carbon transition, OECD

[16] Tong D, Zhang Q, Zheng Y, Caldeira K, Shearer C, Hong C,Qin Y and Davis S J 2019 Committed emissions fromexisting energy infrastructure jeopardize 1.5 ◦C climatetarget Nature 572 373–7

[17] Battiston S, Mandel A, Monasterolo I, Schütze F andVisentin G 2017 A climate stress-test of the financial systemNat. Clim. Change 7 283–8

[18] Sen S and von Schickfus M-T 2020 Climate policy, strandedassets, and investors’ expectations J. Environ. Econ. Manage.100 102277

[19] Binsted M et al 2020 Stranded asset implications of the ParisAgreement in Latin America and the Caribbean Environ. Res.Lett. 15 044026

[20] Iyer G C et al 2015 The contribution of Paris to limit globalwarming to 2 ◦C Environ. Res. Lett. 10 125002

[21] Pfeiffer A, Millar R, Hepburn C and Beinhocker E 2016 The‘2 ◦C capital stock’ for electricity generation: committedcumulative carbon emissions from the electricity generationsector and the transition to a green economy Appl. Energy179 1395–408

[22] Cui R Y et al 2019 Quantifying operational lifetimes for coalpower plants under the Paris goals Nat. Commun.10 4759

[23] Bertram C, Luderer G, Pietzcker R C, Schmid E, Kriegler Eand Edenhofer O 2015 Complementing carbon prices withtechnology policies to keep climate targets within reach Nat.Clim. Change 5 235–9

[24] Platts 2018 World Electric Power Plant Database, Platts[25] Bauer N et al 2017 Shared socio-economic pathways of the

energy sector-quantifying the narratives Glob. Environ.Change 42 316–30

[26] I. E. Agency 2019 Electricity Information 2019, IEA[27] Riahi K et al 2017 The shared socioeconomic pathways and

their energy, land use, and greenhouse gas emissionsimplications: an overview Glob. Environ. Change42 153–68

[28] Tong D et al 2018 Targeted emission reductions from globalsuper-polluting power plant units Nat. Sustain.1 59–68

[29] 2015 IAM helpful or not? Nat. Clim. Change 5 81–81[30] Shearer C, Fofrich R and Davis S J 2017 Future CO2

emissions and electricity generation from proposedcoal-fired power plants in India Earth’s Future5 408–16

[31] Smith P et al 2016 Biophysical and economic limits tonegative CO2 emissions Nat. Clim. Change6 42–50

[32] Matthews H D, Gillett N P, Stott P A and Zickfeld K 2009The proportionality of global warming to cumulative carbonemissions Nature 459 829–33

[33] Allen M R, Frame D J, Huntingford C, Jones C D, Lowe J A,Meinshausen M and Meinshausen N 2009 Warming causedby cumulative carbon emissions towards the trillionth tonneNature 458 1163–6

11

A Self-archived copy inKyoto University Research Information Repository

https://repository.kulib.kyoto-u.ac.jp

Environ. Res. Lett. 15 (2020) 094064 R Fofrich et al

[34] Rogelj J, Forster P M, Kriegler E, Smith C J and Seferian R2019 Estimating and tracking the remaining carbon budgetfor stringent climate targets Nature 571 335–42

[35] Meinshausen M, Meinshausen N, Hare W, Raper S C B,Frieler K, Knutti R, Frame D J and Allen M R 2009Greenhouse-gas emission targets for limiting globalwarming to 2 ◦C Nature 458 1158–62

[36] Raupach M R, Davis S J, Peters G P, Andrew R M, Canadell JG, Ciais P, Friedlingstein P, Jotzo F, van Vuuren D P and LeQuere C 2014 Sharing a quota on cumulative carbonemissions Nat. Clim. Change 4 873–9

[37] Kriegler E, Luderer G, Bauer N, Baumstark L, Fujimori S,Popp A, Rogelj J, Strefler J and van Vuuren D P 2018Pathways limiting warming to 1.5 ◦C: a tale of turning

around in no time? Phil. Trans. R. Soc. A 376 20160457[38] Gerrard M B D and John C 2018 Legal Pathways to Deep

Decarbonization in the United States (Washington, DC:Environmental Law Institute)

[39] Jewell J, Vinichenko V, Nacke L and Cherp A 2019 Prospectsfor powering past coal Nat. Clim. Change 9 592–7

[40] Haley B, Ryan J, Kwok G, Hargreaves J and Farbes J 2019 350PPM Pathways for the United States Deep DecarbonizationPathways Project (DDPP), (San Francisco, CA: EvolvedEnergy Research)

[41] Rauner S, Bauer N, Dirnaichner A, Dingenen R V, Mutel Cand Luderer G 2020 Coal-exit health and environmentaldamage reductions outweigh economic impacts Nat. Clim.Change 10 308–12

12

A Self-archived copy inKyoto University Research Information Repository

https://repository.kulib.kyoto-u.ac.jp