© 2013 by the American Academy of Arts & Sciences

8

A Trillion Tons

Hal Harvey, Franklin M. Orr, Jr. & Clara Vondrich

HAL HARVEY is ceo of EnergyInnovation: Policy and Technologyllc and Senior Fellow for Energyand the Environment at the Paul-son Institute at the University ofChicago.

FRANKLIN M. ORR, JR., is theKeleen and Carlton Beal Professorin the Department of Energy Re-sources Engineering and Directorof the Precourt Institute for Energyat Stanford University.

CLARA VONDRICH is Director of Leadership Initiatives at the ClimateWorks Foundation.

(*See endnotes for complete contributorbiographies.)

How much carbon can humans safely emit intothe atmosphere? Climate scientists argue that a 2degree Celsius (about 4 degree Fahrenheit) increasein global mean temperature is a threshold abovewhich the probability of highly adverse conse-quences grows signi½cantly. Such an increase wouldcorrespond to roughly a trillion tons of total human-caused carbon emissions over time.1 If one trilliontons is humanity’s carbon budget, how much havewe used so far? How fast will we emit the remainderunder current trends? And what can we do to makesure that we don’t bust the budget?To consider the relative contributions of different

variables, ClimateWorks, a foundation that supportspublic policies that mitigate climate change, and its partners at Climate Interactive developed thesystem-dynamics computer model En-roads(Energy–Rapid Overview and Decision-Supportsimulator). En-roads is a global model that assess-es how changes in energy supply and demand mightaffect emissions and, in turn, climate outcomes.2 Itis designed to rapidly assess the impact of variouspolicy scenarios on cumulative emissions by manip-ulating variables as diverse as global gdp, energyef½ciency, innovation, carbon price, and fuel mix.

Abstract: There is a consensus among scientists that stark dangers await in a world where the globalmean temperature rises by more than about 2 degrees Celsius. That threshold corresponds to a collectivehuman carbon emissions “budget” of around a trillion tons, of which half has been spent. This paperuses a new simulation model to look at strategies to stay within that budget, speci½cally assessing theimpact of improvements in energy ef½ciency, aggressive deployment of renewables, and energy technol-ogy innovation. The simulations examine the timing of investments, turnover of capital stock, and theeffect of learning on costs, among other factors. The results indicate that ef½ciency, renewables, and tech-nology innovation are all required to keep humanity within the trillion-ton budget. Even so, these measuresare not by themselves suf½cient: changes in land use and a price on carbon emissions are also needed.

9

Hal Harvey,Franklin M.Orr, Jr. &ClaraVondrich

142 (1) Winter 2013

The simulations underlying this essayemphasize the dynamics of the transitionto clean energy. Changes in energy sys-tems take time because energy productionrelies on large, expensive infrastructurethat is slow to turn over. The oil economy,for example, entails a vast network forexploration, drilling, transport, re½ning,production, automobile manufacture,highway construction, and even human-settlement patterns. Each of these ele-ments cost hundreds of billions of dollarsand took decades to build. The same is truefor coal production and use. To transformthese systems will take decades. The world has spent more than half of

the trillion-ton carbon budget. Currenttrends suggest that, absent policy action,we will exhaust the remaining half byabout 2050.3 The reality of infrastructure“lock-in” and inertia in the global energysupply mix further underscores how vitalit is to consider cumulative carbon emis-sions over time, rather than assessingthem solely on an annual basis. Forinstance, Figure 1 compares a business-as-usual (bau) trajectory with a scenarioin which future emissions remain flat.The ½gure depicts both cumulative emis-sions and annual emissions. In the bau case, emissions increase

annually and sail by the trillionth ton asearly as about 2050. Conversely, one mightexpect that if emissions remain at currentlevels and do not rise, the duration beforewe exhaust the budget would increasesigni½cantly. Yet the flat-emissions sce-nario crosses the trillion-ton line justabout a decade later! In other words, evenif all future demand growth from this pointforward were met by zero-carbon sources, wewould still grossly overshoot the budget. This½nding highlights the reality that it is not enough for us to ramp down emis-sions or keep them flat: we must bringthem to very low levels over the next few decades.

En-roads allows us to investigate dif-ferent options for remaining below thetrillionth ton and gives a sense of howlong it will take for different actions,investments, and policies to reduce emis-sions. The model accounts for the system-wide interactions among different energyoptions. For example, how would the ag-gressive pursuit of energy ef½ciency affectthe growth of renewable energy? How do technology learning curves change thesuite of options? Underlying the model isan extensive study of factors such as con-struction delay times, progress ratios,price sensitivities, historic growth ratesof speci½c energy resources, and energy-ef½ciency potential. Note that En-roads is not a predictive

model; its “results”–which, in this essay,are primarily a calculation of the year inwhich carbon emissions from a given testscenario cross the trillion-ton line–areapproximations only, accompanied by arange of uncertainty. Instead, the modelis a scenario-builder that tests assump-tions about how prospective changes inthe global energy supply mix might affectclimate outcomes.4

If we remain on a bau trajectory, we willsurpass the trillionth ton and the 2 degreeCelsius benchmark–the threshold thatscientists suggest is a dangerous one tocross–around the middle of this century.The good news is that emissions in manyparts of the industrialized world havebegun to flatten and even decline. The badnews is that surging carbon emissions inChina, India, and other rapidly industri-alizing countries do not yet show signs ofabating. By 2030, China’s annual carbonemissions are projected to be around 5 bil-lion tons, compared to around 2.8 billionin 2010. Left unchecked, and in light ofprevailing growth rates, China’s emissionsalone could overwhelm the carbon budgetby the end of the twenty-½rst century.

10

A TrillionTons

Dædalus, the Journal of the American Academy of Arts & Sciences

Figure 1Two Carbon-Emissions Trajectories, 2010–2100

Source: All ½gures created by authors.

Cumulative Carbon Emissions

Annual Carbon Emissions

11

Hal Harvey,Franklin M.Orr, Jr. &ClaraVondrich

142 (1) Winter 2013

Locking in carbon-emitting infrastruc-ture is among the more serious threats ofa bauworld. Commercial and residentialbuildings have useful lives of severaldecades or more. Coal-½red power plantsremain online for up to sixty years.5Newoil and natural gas pipelines could operatefor a half-century. Once these investmentsare made, the economic imperative is touse them. Any new zero-carbon energysource must contend with this embeddedcapital stock. After coal plants are built,their marginal costs of operation are rela-tively small, so the competitive bar fornew technologies is high.The current flood of urban migration in

the developing world offers a useful casestudy of lock-in. China is experiencing thegreatest urban population boom in humanhistory. The United Nations estimates thatChinese cities will add 231 million peopleby 2025 and another 186 million by 20506–numbers roughly equal to the popula-tions of Indonesia and Brazil, respectively.To prepare for this growth, Chinese lead-ers plan to build at least one thousandnew cities.Well-designed cities can slash waste,

reduce air and water pollution, and pro-vide appealing spaces for people to work,shop, and socialize. Poorly designed citiessprawl across the landscape, locking inunsustainable patterns of energy use fordecades. Without policy interventions,including tough building standards andregulations favoring compact develop-ment and low-carbon public transit, baudevelopment is likely to embed a high-consumption pro½le that will be virtuallyimpossible to repair. Indeed, the Interna-tional Energy Agency (iea) has warnedthat without further action, by 2017 allCO2 emissions permitted in its 450 Sce-nario–in which the atmospheric con-centration of carbon dioxide equivalents(CO2e) stabilizes at 450 parts per million(ppm), resulting in an average warming

of 2 degrees Celsius–will be “locked in”by existing power plants, factories, build-ings, and other long-lived infrastructure.7Today’s infrastructure decisions are thusof crucial importance to long-term cli-mate change. Society sets structural pat-terns for future emissions every time ahighway, power plant, factory, or house isbuilt. That capital stock lasts ½fty to onehundred years or more–and every year,contributes carbon to the atmosphere.Furthermore, it is far more costly to repairor retro½t any of these investments thanto get them right in the ½rst place. Theduration of those investments, in light ofthe unforgiving mathematics of carbonaccumulation, means that the windowfor making business-as-usual choices isclosing fast. The iea’s point is that doingthis even for another half-decade locks ina future nobody wants to see.

However mundane it may seem, energyef½ciency is a vital bridge to a low-carbonfuture. Ef½ciency improvements acrossthe transportation, power, and industrysectors would slow or flatten the rise inenergy demand in the coming decades,thereby making the climate problem moresolvable. Aggressive pursuit of energy-ef½ciency solutions on the cutting edge ofengineering, technology innovation, andthermodynamics would yield signi½cantenergy savings. Consider one example: Ifa coal plant is 33 percent ef½cient (theaverage in the United States), and an in-candescent lightbulb is 5 percent ef½cient,then the net conversion of energy to light isabout 1.65 percent. By contrast, a compactfluorescent lightbulb (cfl, roughly 25 per-cent ef½cient), powered by a combined-cycle natural gas turbine (about 60 per-cent ef½cient, using a lower-carbon fuel),converts 15 percent of the energy to light–almost a tenfold increase. The correspond-ing reduction in CO2 emissions per hourof lighting is approximately 90 percent.

12

A TrillionTons

Dædalus, the Journal of the American Academy of Arts & Sciences

The next generation of light-emitting di-ode lamps (leds) will garner even greaterenergy savings.8 Indeed, the Departmentof Energy estimates that replacing regu-lar lightbulbs with leds could potential-ly save 190 terawatt-hours annually, theequivalent of lighting more than 95 mil-lion homes or roughly 5 percent of totalU.S. electricity consumption in 2010.9Similar opportunities exist in virtually

every sector of the economy. Not surpris-ingly, a number of major studies havefound that energy ef½ciency measures area powerful tool for slashing emissions.According to a study from the NationalAcademy of Sciences and the NationalAcademy of Engineering, the technicalabatement potential for energy ef½ciencyin the United States is large; moreover, itscost would be low–or negative, as energysavings would outweigh the costs of newtechnologies over time.10The study foundthat U.S. energy use could fall below bauprojections by 17 to 20 percent in 2020,and by 25 to 31 percent in 2030, providedthat “energy prices are high enough tomotivate investment in energy ef½ciency,or if public policies are put in place thathave the same effect.”11Globally, the prospects for energy ef½-

ciency are bright. A report published byMcKinsey & Company suggests that, withreasonable investments in energy ef½cien-cy, the projected growth in global energydemand could be halved by 2020.12 Thenecessary investments of roughly $170 bil-lion annually would generate an averageinternal rate of return of 17 percent, withtotal energy savings estimated at $900 bil-lion annually by 2020. The investmentstrategy would target only cost-effectiveopportunities, seeking ef½ciency improve-ments across systems such as lighting,cooling, and heating in particular, as wellas vehicles and industrial machinery. The authors of the report caution that

their investment strategy would face a

number of signi½cant challenges. Forinstance, “two-thirds of the investmentopportunity lies in developing countries,where consumers and businesses face avariety of competing demands for theirscarce investment dollars.” In many sec-tors, ef½ciency standards may have to beimplemented to overcome market fail-ures. Nevertheless, the global potential ofenergy ef½ciency to cut carbon at a lowcost is tremendous. What does En-roads tell us about the

effect of aggressively pursuing energyef½ciency? With regard to the globaleconomy, the model’s bau scenario as-sumes an average annual decrease inenergy intensity (energy used per unit ofgdp) of 1.1 percent from 2010 to 2050.13The globally aggregated En-roadsmodeldoes not allow for a manipulation of ef½-ciency improvements across individualsectors or regions. Instead, by varying the energy intensity of gdp, the modelallows us to assess the impact of a highlyef½cient energy economy on the carbonbudget. Figure 2 compares two energy-ef½ciency

scenarios with a bau world: “ef½ciency”(ee) and “ef½ciency plus” (ee+). The eescenario assumes an average annual im-provement in energy intensity of roughly2.5 percent between 2010 and 2050; ee+assumes an average annual improvementof roughly 3 percent. Sustained 2 to 3 per-cent improvements are plausible given abauenergy-intensity improvement rate ofabout 1.1 percent between 2010 and 2050;further, they are consistent with mitiga-tion scenarios examined by other models.In both scenarios, cumulative carbon

emissions still increase but do so moreslowly relative to bau. In the ee scenario,for example, annual emissions in 2050 areabout 42 percent less than in the referencecase. Although annual emissions benddownward in both test scenarios, theyultimately flatten and rise again. Why?

13

Hal Harvey,Franklin M.Orr, Jr. &ClaraVondrich

142 (1) Winter 2013

Figure 2Ef½ciency Test Scenarios, 2010–2100

Cumulative Carbon Emissions

Annual Carbon Emissions

14

A TrillionTons

Dædalus, the Journal of the American Academy of Arts & Sciences

Ef½ciency measures provide large energysavings in the near term, but these gainsare overwhelmed over time if populationand gdp growth continue unabated. The greatest reward in achieving either

of the high-ef½ciency scenarios is thatthey cut the burden for new zero-carbonenergy sources almost in half. Neverthe-less, ef½ciency improvements alone arenot enough to avoid exceeding the budget.The ee scenario reaches the trillionth tonin approximately 2060; ee+ crosses thatline about ½ve years later. While arguablymodest, a ten- to ½fteen-year grace periodprovides needed flexibility in the low-carbon transition and is especially signi½-cant because it can be attained at low cost.

How much can existing and near-termrenewable energy sources, such as solarand wind, help? Renewable energy, ex-cluding hydropower, is the fastest-grow-ing source of electricity generation and isprojected to account for up to a quarter of global electricity generation by 2035(compared to less than 5 percent to-day).14 The eia predicts annual growthrates averaging 3.1 percent between nowand 2035. By contrast, coal-½red electrici-ty generation is expected to grow at anaverage rate of 1.9 percent per year overthe same period.15Recent growth rates are stunning: over

the ½ve-year period from 2005 through2009, global renewable energy capacitygrew at rates of 10 to 60 percent annuallyfor many technologies.16While percent-age increases of this magnitude cannotbe expected to last as the base for renew-ables expands, the forward momentum is undeniable. In 2008, for the ½rst time,more renewable energy than convention-al power capacity was added in both theEuropean Union and the United States.17In 2010, renewables represented half of all newly installed electric capacityworldwide.18

To further put these numbers in con-text, total global power-generating capaci-ty in 2011 was estimated at 5,360 gw. Bythe end of 2011, total renewable powercapacity worldwide exceeded 1,360 gw,up 8 percent over 2010. Renewables thuscomprised more than 25 percent of totalglobal power-generating capacity andsupplied an estimated 20.3 percent ofglobal electricity. Non-hydroelectric re-newables exceeded 390 gw, a 24 percentcapacity increase over 2010. Totalinstalled capacity of non-hydroelectricrenewables in 2010 was around 312 gw,just over 6 percent of the global total. Notethat intermittent renewable energyresources are available only a fraction ofthe time, so nameplate capacity does notdirectly translate into energy. Energy pro-duction is equal to nameplate capacitymultiplied by the capacity factor (thefraction of time plants or installations arein operation), which can vary from about15 percent for solar photovoltaics in cli-mates that do not receive much sunlightto up to 40 percent for offshore windinstallations. Nevertheless, by 2035, re-newables are expected to account forabout a third of global installed capacityand to generate between 15 and 23 percentof the world’s electricity.19Some renewable energy technologies

are close to commercially competitive,including wind. Solar photovoltaics (pv)are approaching the mark, while concen-trated solar thermal power has some dis-tance yet to go. Aggressive deployment ofrenewables can make a big difference rel-ative to the carbon budget. Nevertheless,with the possible exception of onshorewind, these technologies still need to makesubstantial progress along the learningcurve–dropping in price as their volumegrows–to compete with incumbent fossilfuel sources. Evolving technologies, including most

renewable applications, have a high learn-

15

Hal Harvey,Franklin M.Orr, Jr. &ClaraVondrich

142 (1) Winter 2013

ing rate compared to mature technologieslike coal.20 Solar pv has traditionally ex-hibited an average learning rate of 20 per-cent, meaning that price drops by a ½fthfor every doubling of production.21 Ifthese rates persist, a few more doublingsof production capacity could result in costparity with fossil sources. Industry insid-ers suggest that solar pv is on track to bethe cheapest energy source for many partsof the world by the end of the decade.22

There are two caveats, however: solarlearning curves may be flattening23; andthe cost of solar cells is only half of theequation. The enclosure, glass cover,mounting racks, junction boxes, andwiring–together called the balance of sys-tem–are now about half the cost of solarelectricity, and their price decline may beslower.24

Nevertheless, both wind and solar pricescontinue to decline. The solar industry hasachieved manufacturing economies ofscale, and more ef½cient cells are beingdeveloped. The cost of wind energy isalready close to competitive with new gas and coal. According to ren21’s 2011Global Status Report, the cost per kilowatt-hour for onshore wind ranges from 5¢ to9¢, for an average of 7¢/kWh.25

To examine the effect that an accelerat-ed deployment of renewable technolo-gies could have on the carbon budget, wemodeled the following two scenarios:

· Renewables (ren). The cost of energyfrom new renewables was assumed tobe 60 percent below its 2012 price,beginning in the same year. Such a dra-matic decrease in price might comefrom an imagined technological break-through in R&D.26

· Renewables Plus (ren+). A 70 percentdrop in cost was assumed beginning in2012. We also added a 2 percent per yearreduction in the barriers to electrifyingthe transport sector.27

Figure 3 compares emissions in thesetest cases to bau, revealing a substantialdivergence–at least in annual emissions.In 2050, annual ren emissions are 20 per-cent below bau emissions; ren+ emis-sions are 31 percent lower. The differenceis less striking in the cumulative emis-sions view, where ren+ crosses the tril-lionth ton less than a decade after bau. This delay is slight because the dis-

placement of fossil fuels by renewables isminimal in the near decades. Scalingrenewable technologies to meet globalenergy needs remains a challenge evenwhen prices become competitive. Whenrenewables enter the market at or belowthe price of new coal (as they do in bothtest cases), demand for coal declines onlymarginally–a result of the embeddedinfrastructure of fossil-fuel generatingsources. Turning over the capital stock ofcoal plants takes time. Moreover, cost isnot the only factor determining penetra-tion rates of a young technology. Scala-bility and regional differences in windand solar resources, as well as intermit-tency and the corresponding ½rmingrequirements,28 also play a role.

What other technology breakthroughsmight play a signi½cant role in the energytransition ahead? Carbon capture andstorage (ccs) ½gures prominently inmany mitigation scenarios as virtually asine qua non of remaining on a 450 ppmpathway. However, progress has been slowin building the large-scale and capital-intensive demonstration projects neededto test the viability of the technology overits life cycle: that is, from combustion ofthe fossil fuel to the capture and storageof related emissions. In its 2012 report tothe Clean Energy Ministerial, the ieanotes that we can expect to see about tenccs plants operating by the middle ofthis decade. But we will need about 110more by 2020, the agency argues, in order

16

A TrillionTons

Dædalus, the Journal of the American Academy of Arts & Sciences

Figure 3Renewable Energy Scenarios, 2010–2100

Cumulative Carbon Emissions

Annual Carbon Emissions

17

Hal Harvey,Franklin M.Orr, Jr. &ClaraVondrich

142 (1) Winter 2013

to keep global temperature rise below 2degrees Celsius.29

Public ½nancing for ccs peaked in 2008and 2009, when the technology received aboost as part of broader economic stimu-lus programs.30Nevertheless, much of thepromised funding remains unallocated.As of 2012, only 60 percent of the approx-imately $21.4 billion available to supportlarge-scale demonstration projects hadbeen assigned to speci½c ventures.31Ongoing upheaval in the global marketwill undoubtedly continue to squeezeccs budgets. Moreover, ccs is likely toremain underfunded in the absence ofstrong and reliable policy signals, such asa price on carbon. ccs provides just one example of the

potential challenges facing any newbreakthrough technology. Our decades-long experience with solar pv furthercon½rms that, even after invention andinitial deployment, a technology oftenneeds additional help to progress alongits learning curve. Despite being com-mercialized in the 1970s, and notwith-standing its recent and striking pricereductions, solar pv has yet to achieveglobal penetration rates of 1 percent. Cer-tainly, a few more doublings of produc-tion capacity and concomitant pricereductions could revolutionize the out-look for solar, but the sheer scale of thetransition means that this will take time.

Our discussion is by no means a con-demnation of renewables and other newenergy-supply innovations. The aboveexamples merely underscore the intensepressure that innovations face as they tryto gain a foothold in the market. In thecase of renewables, this challenge islargely attributable to the fact that cheapfossil fuels are an entrenched and ubiqui-tous part of our energy economy. Furthercomplicating matters is the uneven play-ing ½eld that results from the failure to

price externalities associated with fossilfuels, including impacts on air quality,human health, and national security,among many other factors. Additionally,renewables are likely to have higher capi-tal costs (but lower operating costs) thanconventional resources, which cost less upfront but require lifetime fuel purchase.Higher capital costs can limit ½nancingfor renewables. Financing also comprisesthe biggest fraction of the levelized cost ofrenewable energy, further undercuttingcompetiveness with established fossilsources. Any other new zero-carbon tech-nology will likely face similar hurdles.But let us adopt a techno-optimist’s

view and assume that a new energy game-changer arrives on the scene. This path-breaking technology circumvents the prob-lems noted above because innovationshave resulted in an energy source so cheapthat the new technology enters the marketat half the price of coal. It is deployable atscale and available for mass penetrationaround the globe. What is the impact ofthis game-changer on the carbon budget? Speci½cally, we de½ned our New Tech-

nology (nt) scenario by the followingassumptions:

· R&D efforts produce a zero-carbon pro-totype in 2020;

· Global deployment takes twelve years;and

· The technology enters the market at halfthe price of new coal.

The nt case assumes that a new zero-carbon energy source, not yet conceived,achieves mass global penetration slightlymore than twenty years from now. Thisambitious scenario exceeds by a wide mar-gin the commercialization trajectories ofany large-scale energy technology existingtoday. For instance, in the renewablessector, wind power has only recentlyachieved signi½cant rates of penetration

18

A TrillionTons

Dædalus, the Journal of the American Academy of Arts & Sciences

in some countries (Denmark, Spain, andGermany) where it has received heavypublic support–more than a half-centuryafter the ½rst grid-connected wind turbinewas manufactured in 1951. Nevertheless,startling innovations are possible. The nt scenario circumvents the most

obvious hurdle to competitiveness–namely, cost–by stipulating that the newzero-carbon energy source enters themarket at half the price of coal. Whilethis strains credulity, it could be possibleif a carbon price were imposed or ifdeployment of the new technology weredeclared a national or international prior-ity. Again, our intent was to test assump-tions at the outer bounds of technical orpolitical feasibility to gauge the impacton the carbon budget. We found that thecarbon emissions accumulated to daterenders the budget relatively insensitiveto even very aggressive assumptions onan energy game-changer. Figure 4 shows that annual nt emis-

sions are 11 percent below bau emissionsin 2050, but crossing the trillionth ton isdelayed by less than a year (though thetest case diverges more substantially frombau as time passes beyond the trillionthton). At ½rst glance, the meager bene½tsof the nt scenario are hard to understand.The assumptions are so radical that onewould expect the needle to move muchmore sharply. In the long run, if a newtechnology is cheaper, we would expect itto take over, partly because of the learningcurve dynamics, which amplify any ini-tial cost advantage (cheaper equals moresales, which leads to more learning andcontinued price reductions). However,this cycle is tempered by capital turnover;even if the new technology is dominantas a share of investment in new capacity,replacing existing capital and achievingdominance overall takes a long time. For En-roads simulations, the frac-

tional investment in various types of

energy supply over the course of eachtime period is determined by the relativeattractiveness of each supply type. Rela-tive attractiveness is a function of thecost of each technology, which in turn isinfluenced by learning, cost of nonrenew-able resources, suitability of remainingsites for renewable energy installations,and capacity for construction of new supply.32 As a result of this structure,market shares of a new technology maynot correspond in the short term withwhat one might expect on the basis ofcost alone.These assumptions about the energy

system imply that coal, oil, and gas willcontinue to be burned for energy through-out the century. Even in scenarios whererenewable energy or new technologiesgrow signi½cantly, drop in price, anddominate the market, fossil fuel energycontinues to be inexpensive (recall thatthere is no price on carbon) and suf½cient-ly available to attract investment and use.Although reliance on fossil fuels is muchless than in the bau scenario, it is stillenough to prolong the increase of cumu-lative emissions.

The En-roads simulations discussedin this essay indicate the value of a diver-si½ed portfolio of emissions-reducingtools. No one tool suf½ces. Ef½ciencyalone, while curtailing demand, cannotstand in for a low-carbon energy source.Renewables are not powerful enough todisplace coal in the near term; we needsustained investments and ef½ciencymeasures to give them time to descendthe learning curve. A new technologybreakthrough, though a crucial long-term solution for global energy needs,requires several decades before it canachieve suf½cient market penetration tomake a difference. Estimated annual emissions do drop

sharply in the individual scenarios. For ex-

19

Hal Harvey,Franklin M.Orr, Jr. &ClaraVondrich

142 (1) Winter 2013

Figure 4New Technology Scenario, 2010–2100

Cumulative Carbon Emissions

Annual Carbon Emissions

20

A TrillionTons

Dædalus, the Journal of the American Academy of Arts & Sciences

ample, in 2050, ee+ emissions are 47 per-cent below bau, and ren+ emissions are31 percent below. Nevertheless, the impacton the cumulative budget is limited. Theaggressive renewables test case buys onlya few additional years before we cross thetrillion-ton line. The nt scenario is alsomarginal, buying less than ½ve yearsbeyond bau. The impact discrepancybetween annual and cumulative emissionsunderscores the inadequacy of nationalemissions reduction schemes based onthe “targets and timetables” approach.For instance, the United States has com-mitted to 80 percent reductions relativeto baseline emissions in 2050. However,unless this target is placed in the contextof cumulative emissions, the numbers arefairly meaningless: “[D]espite makingreference to being guided by the ‘science’,the [Copenhagen] Accord makes nomention of cumulative emissions as thescienti½cally credible framing of mitiga-tion. [Thus] the Accord still falls short ofacknowledging what the science makesabsolutely clear–it is cumulative emis-sions that matter.”33 In other words, wecannot emit willy-nilly until 2049 and thenslash emissions abruptly in 2050 andexpect to be ½ne. Only cumulative emis-sions matter.Nevertheless, we can delay emission of

the trillionth ton much further by deploy-ing a portfolio of actions. For instance, aserious ramp-up of renewables capacitycoupled with an aggressive ef½ciencyportfolio buys more than twenty yearsrelative to bau. Annual emissions in 2050are 57 percent less than bau, and the tril-lionth ton is not emitted until sometimearound 2073. Combining the three mostaggressive scenarios–ee+, ren+, andnt–is marginally better (see Figure 5).34The combined scenario delays crossing

the trillion-ton line by about a quarter-century–a good start, but not suf½cient.Land-use changes are also an essential

piece of the puzzle. Terrestrial sinks, for-ests, and plants have sequestered about aquarter of human-driven carbon emis-sions over time.35 Thus, deforestation inthe Amazon, Indonesia, and the CongoBasin is a major threat, converting a mas-sive carbon-storage sink into a massivecarbon source. The draining and burningof peat bogs is another major globalsource of CO2 emissions–indeed, thethird largest after burning fossil fuels and deforestation. Unsustainable farmingpractices are also to blame. For instance,the carbon-rich grasslands and forests intemperate zones have been replaced bycrops with a much lower capacity tosequester carbon. Aggressive policies areneeded to arrest these developments andfurther forestall the trillionth ton. The last missing piece is the interaction

of economic growth and coal. The sce-narios modeled in this essay reflect thisrelationship in the second half of the cen-tury, when carbon emissions begin togrow after a period of steady decline. Inthe ee+ and ren+ scenarios, for example,annual emissions decline through themiddle of the century only to rise again.Why does this happen? Quite simply, theclean energy supply is overwhelmed bygrowth, absent additional downwardpressure on coal. If gdp continues to growby 2 to 3 percent a year and coal remainscheap, ef½ciency gains and renewableswill not keep pace with demand.36A priceon carbon or an international deal onemissions reductions could alter this pic-ture. Though policy options to achieve anordered reduction of coal are manifold, weuse the proxy of a carbon price in our ½nalscenario below. Our ½nal scenario combines ee+, ren+,

and ntwith two new actions or policies:a CO2price of $35 per ton starting in 2025,and a 50 percent reduction in emissionsfrom land-use sectors and other moreshort-lived greenhouse gases.

21

Hal Harvey,Franklin M.Orr, Jr. &ClaraVondrich

142 (1) Winter 2013

Figure 5Combination Scenarios, 2010–2100

Cumulative Carbon Emissions

Annual Carbon Emissions

22

A TrillionTons

Dædalus, the Journal of the American Academy of Arts & Sciences

As Figure 5 shows, this scenario (“theSuite”) holds cumulative emissions belowthe trillionth ton throughout this century.In 2050, annual emissions are 67 percentbelow bau. Speci½c components of theSuite could be adjusted, of course. Use ofnuclear power (in place of coal) couldreduce emissions associated with base-load electric power generation, and recentchanges in the availability of natural gascould also be explored. A more stringentland-use policy might be traded for a car-bon price. This thought experiment sim-ply highlights that there is no single solu-tion: multiple measures are needed tokeep humanity on a reasonable climatetrajectory.

The En-roadsmodeling exercise showshow rapid deployment of ef½ciency im-provements, renewables, and new tech-nologies might impact the carbon budgetover this century. Speci½cally, it con½rmsthat each component is necessary andnone is suf½cient alone. Combined with acarbon price and effective land-use poli-cy, these three tools offer a challengingbut credible path that stays within the

carbon budget. We do not have time towaste if we are to avoid dangerous, irre-versible climate changes for which mod-ern civilization is ill-suited to adapt.Indeed, the iea warns that we have ½veyears to get off the bau path. After thatpoint, the energy infrastructure we buildwill start to lock in emissions-generatinginfrastructure that will push globalwarming beyond 2 degrees Celsius. Starting today, we can use existing

tools to begin a steady decline in emis-sions at low cost. In coming years, renew-able power will grow cheaper, while fossilfuel prices will adjust to future suppliesand competition from other resources. Abreakthrough innovation may well revo-lutionize the energy sector in ways we cannow only dream about. Programs andpolicies to foster advances can be pursuedon a national, state, or local level; policy-makers and businesses at every level areempowered to take action now. The sce-narios examined here suggest that withan aggressive and sustained effort, we canpush back the timetable for expending ourcarbon budget and thus sharply reducethe risks of surpassing the trillionth ton.

endnotes* Contributor Biographies: HAL HARVEY is ceo of Energy Innovation: Policy and Technol-

ogy llc and Senior Fellow for Energy and the Environment at the Paulson Institute at theUniversity of Chicago. He is the founder of the ClimateWorks Foundation, where he previ-ously served as ceo. He is former Environment Program Director at The William and FloraHewlett Foundation and founder and past President of the Energy Foundation. An energyengineer, he has worked on energy policy in a dozen countries.

FRANKLIN M. ORR, JR., is the Keleen and Carlton Beal Professor in the Department of Ener-gy Resources Engineering and Director of the Precourt Institute for Energy at Stanford Uni-versity. He is also a Senior Fellow at the Woods Institute for the Environment. His researchfocuses on understanding the physical mechanisms that control displacement performancein gas injection processes for oil recovery and storage of greenhouse gases. His work hasappeared in the Journal of Fluid Mechanics, Energy Conversion and Management, and the Inter-national Journal of Greenhouse Gas Control, among other publications.

CLARA VONDRICH is Director of Leadership Initiatives at the ClimateWorks Foundation.She is an attorney with a background in international arbitration and energy policy. Previ-ously, she served as Counsel to the President’s Commission on the bp Deepwater HorizonOil Spill, focusing on ecological restoration and environmental justice. She also helped pre-

23

Hal Harvey,Franklin M.Orr, Jr. &ClaraVondrich

142 (1) Winter 2013

pare amicus briefs in support of State plaintiffs in the landmark U.S. Supreme Court case,American Electric Power v. Connecticut, which sought to hold major electric utilities liable fortheir contribution to climate change.

Acknowledgments: We would like to thank Andrew Jones and Elizabeth Sawin of ClimateInteractive for their unfailing support and collaboration at all stages of producing this paper.

1 Myles Allen et al., “Warming Caused by Cumulative Carbon Emissions Towards the Tril-lionth Tonne,” Nature 458 (April 30, 2009): 1163–1166. Other carbon budgets have been pro-posed, depending on the “acceptable risk” stipulated for crossing the 2 degree Celsiusthreshold. For instance, Malte Meinshausen and colleagues propose a carbon budget ofroughly 830 billion tons to assure a 50 percent chance of staying under 2 degrees Celsius. SeeMeinshausen et al., “Greenhouse-Gas Emission Targets for Limiting Global Warming to 2° C,”Nature 458 (April 30, 2009): 1158–1162. Despite these differences, the “need in principle fora cumulative limit near or below one trillion tonnes is generally accepted”; see Myles Allenet al., “The Case for Mandatory Sequestration,” Nature Geoscience 2 (2009): 813–814.

2 En-roads was created by Climate Interactive, Ventana Systems, mit Sloan School of Man-agement, and the ClimateWorks Foundation global research team. It is designed to comple-ment, not supplant, other more disaggregated models addressing similar questions, such asthose in the Energy Modeling Forum’s emf-22 suite. En-roads relies on other models andEnergy Information Administration (eia) projections for testing and data. It is based on thePh.D. dissertations of John Sterman, Professor in the mit Sloan School of Management, andTom Fiddaman of Ventana Systems. We wish to distinguish En-roads from the mit Emis-sions Prediction and Policy Analysis model (eppa), another simulation related to climateand energy. eppa was developed as part of mit’s Joint Program on the Science and Policy ofGlobal Change. En-roads is a system-dynamics (high-order nonlinear differential equa-tion) simulation; a more detailed description of En-roads methodology and assumptionsis available at http://climateinteractive.org/simulations/en-roads/en-roads.

3 Allen et al., “Warming Caused by Cumulative Carbon Emissions Towards the TrillionthTonne.” See also Catherine Brahic, “Humanity’s Carbon Budget Set at One Trillion Tonnes,”New Scientist, April 29, 2009. But compare Meinshausen et al., “Greenhouse-Gas EmissionTargets for Limiting Global Warming to 2° C.”

4 The En-roads model does not include full economic feedbacks. For example, it does notcapture the effect of energy prices on the aggregate economy’s growth rate, or the effect ofclimate impacts on economic growth. The model is not suitable, therefore, for exploringoptimal trade-offs between mitigation and climate adaptation. Further, the current versionof En-roads does not explicitly include carbon capture and storage, the dynamics of invest-ment in and payoff from research and development, energy storage technologies, popula-tion, or labor. But it does include surrogates for each of these variables–namely, in its abil-ity to model generic new clean-energy technologies, technology learning curves, and gdpgrowth.

5 Estimates of the “useful life” of a coal plant vary from thirty to sixty years. Concerns aboutthe conventional pollutants emitted by coal-½red power plants (sulfur and nitrogen oxides,particulate matter, and mercury) have led to increasing pressure in some countries to shutdown the worst offenders. Still, the large base of coal plants and the low marginal cost ofoperation mean that many will run for a long time.

6 United Nations Department of Economic and Social Affairs/Population Division, WorldUrbanization Prospects: The 2009 Revision (New York: United Nations, 2010), 12.

7 World Energy Outlook 2011 (Paris: International Energy Agency, 2011). 8 At the moment, cfls are more ef½cient than leds. Current led ef½cacy is around 50 lumens

per watt (60 lm/W max for some high-end Japanese products), whereas cfls reach, at themost, 60 to 70 lm/W. cfls will likely remain the better choice for general lighting for another½ve to ten years; leds are more likely to be used for specialty applications at ½rst.

24

A TrillionTons

Dædalus, the Journal of the American Academy of Arts & Sciences

9 America’s Energy Future Panel on Energy Ef½ciency Technologies, Real Prospects for EnergyEf½ciency in the United States (Washington, D.C.: National Academies Press, 2010).

10 Ibid. 11 Ibid.12 Diana Farrell and Jaana K. Remes, “How the World Should Invest in Energy Ef½ciency,”

McKinsey Quarterly (July 2008).13 The bau scenario includes an annual improvement in primary energy intensity of 1.1 percent

per year from 2010 to 2050. This estimate is somewhat lower than the reference scenariosdeveloped by the iea (approximately 1.8 percent per year from 2010 to 2050) and the eia (1.7percent per year from 2007 to 2035); but it falls in the range of emf-22 reference scenarios(0.4 to 1.5 percent per year). See World Energy Outlook 2011; Energy Information Adminis-tration, International Energy Outlook 2010 (Washington, D.C.: U.S. Department of Energy,2010), Appendix J, “Kaya Identity Projections”; Leon Clarke et al., “International ClimatePolicy Architectures: Overview of the emf 22 International Scenarios,” Energy Economics 31(2009): S64–S81.

14 Energy Information Administration, International Energy Outlook 2011 (Washington, D.C.:U.S. Department of Energy, 2011), 86, Table 11. When hydropower is included, renewablesources already supply nearly 20 percent of global electricity generation. See ren21, Renew-ables 2011: Global Status Report (Paris: ren21 Secretariat, 2011).

15 International Energy Outlook 2011, 86. Note that coal-½red generation, which starts from a farhigher base than renewables, still experiences cumulative growth of 67 percent during thattime: from 7.7 trillion kilowatt-hours in 2008 to 12.9 trillion kWh in 2035.

16 ren21, Renewables 2010: Global Status Report (Paris: ren21 Secretariat, 2010), 15.17 ren21, Renewables Global Status Report: 2009 Update (Paris: ren21 Secretariat, 2009), 8.18 ren21, Renewables 2011: Global Status Report.19 Compare World Energy Outlook 2011, 178. The iea’s report forecasts the share of generation

from non-hydroelectric renewables to be 15 percent in 2035, under its “New Policies Sce-nario.” Also compare International Energy Outlook 2011, 86, Table 11. The eia’s report forecastsnet electricity generation of renewables to be more than 23 percent.

20 Tooraj Jamasb, “Technical Change Theory and Learning Curves: Patterns of Progress in Ener-gy Technologies,” The Energy Journal 28 (3) (2007): 51–71.

21 See, for example, Technology Roadmap: Solar Photovoltaic Energy (Paris: International EnergyAgency, 2010), 18. The iea notes that pv module costs have decreased at historical learningrates of 15 to 22 percent and, further, that balance-of-system cost reductions have kept pace.The iea assumes a prospective learning rate of 18 percent for the whole pv system on thatbasis.

22 Kees van der Leun, “Solar pv Rapidly Becoming the Cheapest Option to Generate Electric-ity,” Grist, October 11, 2011, http://www.grist.org/solar-power/2011-10-11-solar-pv-rapidly-becoming-cheapest-option-generate-electricity.

23 David Keith and Juan Moreno-Cruz, “Is the Solar Photovoltaic Learning Curve Flattening?”Near Zero Technical Note, June 15, 2011, http://nearzero.org.

24 The extent to which balance-of-system costs are falling is an open question, with some partiesquite optimistic. For instance, the iea maintained that balance-of-system cost reductionshave kept pace with the historical solar panel learning rates of roughly 20 percent. See Tech-nology Roadmap: Solar Photovoltaic Energy, 18.

25 ren21, Renewables 2011: Global Status Report, 33 at Table 1. (“All costs in this table are indica-tive economic costs, levelized, and exclusive of subsidies or policy incentives. Typical energycosts are under best conditions, including system design, siting, and resource availability.

25

Hal Harvey,Franklin M.Orr, Jr. &ClaraVondrich

142 (1) Winter 2013

Optimal conditions can yield lower costs, and less favorable conditions can yield substan-tially higher costs.”)

26 A 60 percent cost reduction in renewable energy is clearly extreme. We sought to push thelimits of technical feasibility for each test case to explore the ultimate impact on the carbonbudget. That even extreme assumptions, individually or in partial combination, failed toprevent crossing of the trillionth ton is sobering.

27 Changes in investment and policy that would make it easier to electrify transportation weremodeled by removing barriers to smarter grids, better electric vehicles, and complementaryinfrastructure such as fueling, parts, and maintenance.

28 Firming refers to use of a dispatchable backup resource (hydroelectric power or natural gas)to supplement an intermittent resource in order to ensure that energy supply is suf½cient tomeet demand.

29 Tracking Clean Energy Progress: Energy Technology Perspectives 2012: Excerpt as IEA Input to theClean Energy Ministerial (Paris: International Energy Agency, 2010). The iea scenario suggeststhat we need to capture about 270 million tonnes (Mt) CO2 from the power and industrysectors in 2020. Meanwhile, the Global ccs Institute notes that the CO2 storage capacity ofthe fourteen large-scale projects currently in operation or under construction is only about33 Mt CO2 per year.

30 For instance, President Obama allocated $3.4 billion to ccs research development and deploy-ment programs as part of the American Recovery and Reinvestment Act in 2009. Meanwhile,Europe established a ½nancing program under the Emissions Trading Scheme to supportlarge-scale demonstration projects for ccs and other low-carbon energy technologies (the“ner 300” program, so called because it allocates 300 million allowances in the New Entrants’Reserve of the eu Emissions Trading Scheme).

31 The Global Status of CCS: 2011 (Canberra, Australia: Global ccs Institute, 2011).32 For details, see http://climateinteractive.org/simulations/en-roads/en-roads.33 Kevin Anderson and Alice Bows, “Beyond ‘Dangerous’ Climate Change: Emissions Scenarios

for a New World,” Philosophical Transactions of the Royal Society A 369 (1934) (January 13, 2011):20–44.

34 This result is partially explained by market competition. A driving force in the scale-up oftechnologies is the “learning by doing” effect, whereby prices fall and investment becomesmore attractive as installed capacity increases. When new technology and renewables growsimultaneously, neither accumulates an installed total as quickly.

35 Yude Pan et al., “A Large and Persistent Carbon Sink in the World’s Forests,” Science 333(6045) (2011): 988–993.

36 These growth rates are used in En-roads simulations (and those in the emf-22 suite, forexample). Whether they are realistic for the end of the century is an open question.