Potential Early Markets for Fusion Energy
JJ. Fusion Energy manuscript No. (will be inserted by the editor)
Potential Early Markets for Fusion Energy
Malcolm C. Handley* · Daniel Slesinski † · Scott C. Hsu
Received: date / Accepted: date
Abstract
We identify potential early markets for fu-sion energy and their projected cost targets, based onanalysis and synthesis of many relevant, recent studiesand reports. Because private fusion companies aspire tostart commercial deployment before 2040, we considerpotential markets for fusion in 2035, including electric-ity, process heat, and hydrogen production. We vari-ously consider business-as-usual” and high-renewables-penetration scenarios, as well as carbon pricing up to100 $ /tCO . Key findings are that fusion developersshould focus initially on high-priced global electricitymarkets and include integrated thermal storage in or-der to maximize revenue and compete in markets withhigh renewables penetration. Process heat and hydro-gen production will be tough early markets for fusion,but may open up to fusion as markets evolve and if fu-sion’s levelized cost of electricity falls below 50 $ /MWh e .Finally, we discuss potential ways for a fusion plant toincrease revenue via cogeneration (e.g., desalination, di-rect air capture, or district heating) and to lower capitalcosts (e.g., by minimizing construction times and inter-est or by retrofitting coal plants). Keywords
Fusion energy · Energy markets
Supported by DOE ARPA-E, this is the work of the authorsand does not necessarily represent the views of ARPA-E,DOE, or the U.S. government.Corresponding author: Scott C. HsuE-mail: [email protected] Research Projects Agency-EnergyU.S. Department of EnergyWashington, DC 20585, USA*Malcolm C. Handley is now unaffiliated † Daniel Slesinski is now with Johns Hopkins University
It is widely assumed that when fusion energy gain isdemonstrated, humanity will be on the cusp of an ageof economical, abundant, and carbon-free energy. In-deed, there are reasons to believe that if fusion poweris achieved and allowed to mature, low costs mightfollow. Assuming that technical feasibility is demon-strated, several things must happen for fusion to reachmaturity: regulations must not stifle it; the public mustaccept it; and governments and investors must supportit. Most importantly, fusion must find early marketsthat are large and profitable enough to continue to de-velop. Contributing to deep decarbonization may thenfollow. In this paper, we identify fusion’s potential earlymarkets and their cost targets.Many private fusion companies aspire to start com-mercial deployment before 2040. Therefore, we examinepotential markets for fusion in 2035, including electric-ity, process heat, and hydrogen production. We vari-ously consider scenarios with business as usual,” highpenetration of variable renewables on the grid, and car-bon pricing up to 100 $ /tCO . We do not considercarbon pricing > $ /tCO because direct-air car-bon capture and sequestration is likely to cost less. Forpower generation, we consider only grid scale becausesmaller ones are primarily targeting small, off-grid mar-kets that appear unlikely to become a substantial en-ergy market.Our study is conservative. We ignore markets thatrequire large changes in the energy sector or existinginfrastructure because these add to the many commer-cialization risks that fusion already faces. We take asomewhat unforgiving view of the competitive situa-tion. We assume that other energy technologies willcontinue seeing large cost reductions, and we do not a r X i v : . [ phy s i c s . s o c - ph ] J a n M. C. Handley, D. Slesinski, and S. C. Hsu explicitly take credit for less-tangible benefits of fusion,such as the potential for being sited near populationcenters and lower regulatory risks compared to nuclearfission. We caution the reader to keep these caveats inmind. If and when fusion is proven at scale with rea-sonable costs (even if initially higher than the compe-tition), it is likely to be a disruptive energy technologythat may fundamentally and eventually alter marketsand the way humans use energy. Fusion may well de-liver on the promise of clean, abundant, safe, affordableenergy and replace many other energy sources. Its long-term markets are potentially enormous.Researched and written in support of the Tech-to-Market (T2M) component of the ARPA-E fusion port-folio [1,2,3,4], this paper is intended to be an informa-tional resource both for ARPA-E fusion performers andfor potential fusion investors, to help them focus on ini-tial markets that can nurture fusion through its earlydeployments. It is also for the broader fusion researchcommunity, to whom we hope to clarify the character-istics that fusion may need, at least initially, in orderto succeed in the marketplace. While fusion is the mainfocus of this report, much of the analysis is relevant toadvanced nuclear fission.Finally, regulation and licensing will clearly impactfusion development, cost, and time to market. Fusionregulation is an important subject that is outside thescope of this paper. We refer the reader to recent whitepapers [5,6,7,8] and presentations [9] on the subject.The paper is organized as follows. The next section,“Key Findings,” is an executive summary of the pa-per, followed by sections that detail the key findingson “Electricity,” “Process Heat,” “Hydrogen and ItsDerivatives,” and “Economic Boosts.” The paper closeswith a “Summary and Conclusions.”
Energy markets are fiercely competitive and have lowmargins. However, there are many specific markets withhigher energy prices that can serve as beachheads forfusion. Key findings of this paper are summarized inthis section. Details, analysis, and references supportingthese findings are given in the following sections.
Electricity is a promising initial market and can po-tentially support fusion with levelized cost of electricity(LCOE) as high as ∼ $ /MWh e in some global re-gions. Electricity is a commodity, which means that spe-cific offtake partners do not need to be found for eachproject. It can be traded, unlike process heat, thoughnot globally, and hence is subject to significant regional price variation. This variation is driven by different fuelcosts and, increasingly, different levels of renewablespenetration and carbon pricing, and might make somemarkets initially more viable/profitable for fusion. Wefind that there could potentially be a very large marketfor fusion power plants with an LCOE of 50 $ /MWh e . Integrated thermal storage will help fusion compete onpower grids dominated by solar and wind, which bid at0 $ /MWh e for much of the day. Thus, fusion must beable to deliver a competitive LCOE even while sellingelectricity for as little as 12 hours per day. Because ofthe high capital and fixed costs and low variable costs ofa fusion power plant, a capacity factor of 0.5 would seta high LCOE. Onsite thermal storage allows the fusioncore to run continuously while selling power only whenthe grid price is high, enabling lower overall LCOE.Thermal storage appears to be a low-cost addition toa fusion power plant. In these situations, direct gener-ation of electricity from fusion (i.e., without heat as anintermediary) may be less economically beneficial thanpreviously expected. Process heat may be a difficult early market for fusion.It is a hyper-localized market to which fusion powerplants must be tailored. The fusion power plant wouldlikely need to be sited next to facilities that utilizethe heat. These constraints result in slower and riskierdeployment of new heat sources. Most of the marketwill initially be inaccessible to fusion because of high-temperature requirements, the fulfillment of which byfusion will require additional materials research and de-velopment. In addition, because many heat-generationprocesses run off fuel that is produced as a byproductof the process, cost targets are likely to be below reachfor early deployments of fusion.
Hydrogen production may also be a difficult early mar-ket for fusion. Hydrogen prices vary less than electricityprices but perhaps by enough to create some potentiallyviable early markets. For example, if fusion can get to78 $ /MWh e or below, then it may become feasible touse fusion to produce hydrogen as a partial substitutefor natural-gas heating of buildings. Additional revenue sources can improve the economicsof fusion power plants. Cogeneration or use of wasteheat from a fusion plant to power direct-air carbon-capture equipment, desalination equipment, or district-heating networks can significantly improve the overalleconomics. Modeling suggests that the effective cost ofenergy can be lowered by as much as 35% in some cases. otential Early Markets for Fusion Energy 3
Retrofitting/repowering coal plants might save 30% ofthe capital cost of fusion power plants by reusing manyof the balance-of-plant components.
Carbon pricing will help fusion relative to the fossil-fuelcompetition. Although there are many markets wherefusion could be competitive without carbon pricing,the latter expands the markets for fusion. One fifth ofthe worlds greenhouse-gas emissions are already cov-ered by a carbon price, ranging from insignificant toover 100 $ /tCO in Scandinavia [10, p. 15]. A priceof ≥ $ /tCO could allow fusion to be competitivewith fossil fuels almost everywhere if fusion costs ≤ $ /MWh e . The impact of carbon pricing on the hy-drogen market is less significant. Grid electricity is a promising initial market for fu-sion because electricity is a commodity and has sig-nificant price variations, including some regions withmuch higher prices. Furthermore, a fusion power plantwill not require customization for a particular customer.Electricity is traded enough that a power plant is notdependent on one nearby customer, but not traded somuch that a few ultra-low-cost players dominate theglobal market. Fusion may be able to penetrate mostmarkets if it can eventually reach costs of ≤ $ /MWh e .Beyond the factor of cost, areas best suited for the firstfusion power plants may be those with higher popula-tion densities and lower land availability/suitability forlarge-scale renewables generation.3.1 Early MarketsTable 1 shows a sampling of wholesale electricity pricesfrom around the world. Because the first fusion powerplants will likely be more expensive than most incum-bent forms of power, specific high-priced markets maybe best suited for early fusion deployments.Singapore could be one such location, where in 2018the average wholesale electricity price was 110 $ /MWh e ,with 95% coming from natural gas. In addition, beingone of the most densely populated areas of the world,Singapore will need an energy source that can be lo-cated near or within the city, while delivering largeamounts of power with little land usage.With a wholesale electricity price of 92 $ /MWh e andwith limited land for renewable energy, Japan may beanother favorable country for early fusion deployment.In addition, Japan has a high level of technological ca-pability to facilitate rapid fusion deployment. Table 1
Regional wholesale electricity prices [11, Tab: Globalwholesale prices].Market Benchmark Price Market( $ /MWh e ) (GW)Singapore Average wholesaleprice 110 6Japan Average wholesaleprice 92 108U.S. Northern CA whole-sale price 61 38Poland Baseload wholesaleprice 60 17Italy Baseload wholesaleprice 59 33U.K. Baseload wholesaleprice 55 35Slovenia Baseload wholesaleprice 55 2Portugal Baseload wholesaleprice 54 5Spain Baseload wholesaleprice 53 27Estonia Baseload wholesaleprice 51 1U.S. Average wholesaleprice 39 445World Power from gas,coal, and nuclear(2035) – 2283 In the U.S., northern California has a wholesaleelectricity price of 61 $ /MWh e , which is considerablyhigher than the rest of the country. New England is acandidate early market as well because it is not idealfor onshore wind nor solar additions due to a lack ofsunny days and large swaths of open land. In addition,the northeast as a whole has the highest populationdensity in the country, with many large, power-hungrycities.Much of Europe could become available if fusioncosts ≤ $ /MWh e .3.2 Capacity PaymentsUsing storage and transmission to provide reliable powerfrom variable renewables can add considerable costs.As a result, many U.S. electricity markets offer capac-ity payments to firm (i.e., not variable) power plantsto encourage their construction. These payments are75 $ /kW e per year in New England and have recentlybeen 100 $ /kW e per year on the PJM grid ( ) [12, p. 33]. At these lev-els, such payments can make a significant contributionto the economics of a fusion power plant, reducing theLCOE by 10 and 13 $ /MWh e , respectively [11, Tab: Ca-pacity Payments]. Thus, choosing a location with high M. C. Handley, D. Slesinski, and S. C. Hsu and predictable capacity payments could significantlyimprove the economics of a fusion power plant.3.3 High RenewablesGrids with high renewables penetration present a chal-lenge for fusion because the marginal cost of highly in-termittent renewables production is zero. The result isthat whenever renewables are producing power, theymeet the entire grid demand and are likely the lowest-cost power source available, but there are long periodswhen renewables produce little or no power.
Integrated thermal storage can help make fusion com-petitive on grids with significant renewable penetration.The traditional response to variable, low-price powerfrom renewables is to ramp down other power plants(e.g., natural-gas plants) during times of high renew-ables production. However, unlike natural-gas plants, afusion power plant will have high capital and fixed costsrelative to its fuel costs, meaning that little money issaved when a fusion plant idles.If idling the plant saves no money, then idling theplant for half of the time will double the LCOE. Forexample, consider a hypothetical future fusion plantthat faces a similar situation as the example of a fis-sion power plant in southern California with a capacityfactor of only 0.67 in a high-renewables scenario [12,p. 42]. Table 2 illustrates the economics of such a sce-nario. The Default plant column describes a plant with0.92 capacity factor and LCOE of 41 $ /MWh. As thecapacity factor drops to 0.67 and then 0.5 because ofvariable, low-cost renewables, the plant LCOE rises by36% and then 82% relative to the baseline, respectively.However, if a hypothetical fusion plant can divertits output to storage, then the fusion core can continueto run at high capacity factor instead of idling dur-ing periods of high production by variable renewables.When the grid price for electricity rises, the plant cansell stored energy as well as the energy that it contin-ues to generate. For example, a fusion core might powera plant that delivers, for limited periods, a peak powerthat is double that of the fusion core alone, with the dif-ference coming from thermal storage. Thus, the sameplant fitted with thermal energy storage performs muchbetter in the face of high renewables penetration. Suffi-cient storage to cover 10–12 hours of production couldallow for all-day operation of the fusion core.A National Renewable Energy Laboratory (NREL)cost model for concentrated solar power with molten Fig. 1 “Duck curve” with fusion and thermal storage. Figureis adapted from ISO New England. salts [13] suggests that ten hours of storage for a 400-MW e plant would cost $ Using curtailed fusion electricity to generate hydrogenwhen renewables are over-producing does not appear tobe economically viable. During these times, the excesspower from renewables is unused and available for pur-chase. Any consumer would do better to buy renewablepower instead of more expensive fusion power.3.4 Accessing Larger MarketsThe ideal outcome for fusion in the longer term is tohave a lower LCOE than the competition, as has beenthe case recently with natural gas costing less than coal. otential Early Markets for Fusion Energy 5
Table 2
Impact of thermal storage [11, Tab: Storage].Type Item Units Default plant High renewables Higher renewables With storageInput Rated power MW 400 400 400 400Input Capacity factor % uptime 92 67 50 92Fusion capex – $ /MW 3,000,000 3,000,000 3,000,000 3,000,000Storage capex – $ /MWh – – – 35,000Ten-hour stor-age capex – $ /MW – – – 350,000Capex total – $ /MW 3,000,000 3,000,000 3,000,000 3,350,000Opex Fixed O&M $ /MWyr 10,800 10,800 10,800 12,060Opex Variable O&M $ /MWh 1.20 1.20 1.20 1.34Opex Fuel (assume free) $ /MWh 0 0 0 0Output LCOE $ /MWh 41.14 56.04 74.69 45.94Output LCOE relative todefault % 100 136 182 112 In this subsection, we examine market prices as a func-tion of the prices of natural gas and coal, as well as thepenetration of renewables. We find that for fusion tobe competitive in most markets, its LCOE should be inthe range 40–50 $ /MWh e .For fusion to be competitive in many large markets,it may need to reach a similar LCOE to that of natu-ral gas. While the LCOE of natural-gas generation willvary based on the fuel cost, as examined below, it isuseful to consider present LCOE estimates of naturalgas, which have a range of 44–68 $ /MWh e [14, p. 2],as a benchmark. For new natural gas coming into ser-vice in 2025, the EIA reports LCOE of 38 $ /MWh e fornatural gas combined cycle (NGCC) and 67 $ /MWh e for combustion turbine [15, p. 7]. This shows that anaggressive benchmark for the cost of fusion would beabout 40 $ /MWh e .While the above costs are a useful reference, it isimportant to further note that the price for natural gaswill vary based on location, with the fuel being a sig-nificant portion of the total cost. A study from Google,which estimates the cost of integrating intermittent anddispatchable electricity sources on a grid [16], finds thata fusion-like (i.e., hypothetical firm, low-carbon energysource) power plant would undercut all other energysources in the U.S. market, including existing plants,if it could produce electricity at 17 $ /MWh e even whennatural gas costs as little as 2.5 $ /MMBTU. This wouldpotentially open up a 308-GW market to fusion plantsas fast as they can be built. The Google study alsofinds that with a less-aggressive LCOE of 28 $ /MWh e ,a fusion-like system could replace existing power plantsas they are retired. This would create a demand of 12GW of new fusion plants per year. See [11, Tab: Replac-ing existing assets] for calculations of these numbers.Regional variation in the price of electricity is drivenby variation in fuel costs, carbon price, and renewablepotential. Using cost assumptions from the U.S. mar- Fig. 2
LCOE for NGCC generation for various carbon andfuel prices [11, Tab: LCOE for NGCC]. ket, we examine the price of competing power sourcesas a function of these inputs [16]. Figure 2 relates theprice of natural-gas generation to the price of fuel, thecarbon price, and whether carbon capture and seques-tration (CCS) is used. Natural-gas prices in 2019 wereabout 2.5 $ /MMBTU in the U.S., with the highestbeing 10 $ /MMBTU in Japan; only in the U.S. andCanada are prices below 4.4 $ /MMBTU [17, p. 39].Figure 2 suggest that fusion at (cid:46) $ /MWh e can po-tentially become competitive with natural gas over arange of projected scenarios.The final firm power source we consider as a com-petitor to fusion is coal. In Fig. 3, we relate the price ofelectricity from coal power plants to the price of coal,the carbon price, and whether CCS is used. Over thelast 20 years, coal prices have been as low as 4 $ /MWh t and as high as 20 $ /MWh t [18], ignoring the price forcoking coal. In 2018, prices ranged from 9 $ /MWh t inthe U.S. to 11 $ /MWh t in Europe to 14 $ /MWh t inAsia. These data suggest that fusion at (cid:46) $ /MWh e can potentially outcompete coal over a range of pro-jected scenarios. M. C. Handley, D. Slesinski, and S. C. Hsu
Fig. 3
LCOE for coal generation for various carbon and fuelprices [11, Tab: LCOE for Coal].
Fig. 4
LCOE versus renewable fraction in California [11,Tab: SCOE for renewables].
Wind and solar produce very low-cost power, butthe latter is not always the same as a low system-wideLCOE. Rising penetration of renewables leads to risingmarginal costs because of the mismatch between thetimes and locations of supply and demand, and the costof transporting and storing energy, as shown in Fig. 4for California. The future capital and fixed costs for so-lar are 0.63 $ /W and 21.66 $ /MWh e , respectively, and1.37 $ /W and 46.71 $ /MWh e , respectively, for wind[18]. Recent work has shown that including a firm, low-carbon source (like fusion) would reduce overall systemcost in deep-decarbonization scenarios for power gener-ation [19].Based on the analyses of this subsection, we con-clude that fusion with an LCOE of 41 $ /MWh e shouldbe able to compete with natural gas in most countries,coal in almost all countries, and renewables when theymake up 35% or more of the grid. With a carbon price ofat least 20 $ /tCO , fusion at 50 $ /MWh e can competewith gas in most countries other than the U.S., coalin all countries, and renewables when they are 50% ormore of the grid. If fusion can reach a low enough price to compete with natural gas, it could be an excellentcomplement to high levels of renewables. By 2035 andbeyond, more regions will likely have high penetrationof renewables, and these could be good early marketsfor fusion. Selling to countries with the largest growth in electric-ity demand may require providing modular equipmentor using a build-own-operate (BOO) model. Buildinglarge infrastructure projects anywhere is risky, and suchprojects often go over their original budget and/or sched-ule. Considering energy factors such as how much cleanenergy a country already has, economic factors suchas GDP, and institutional factors such as regulatoryquality, a recent study [20] finds that 85% of demandgrowth will be in countries with below-median economicand institutional readiness. Fusion will find more suc-cess in these markets, and more rapidly, if plants aresmall and modular, so that their output can be moreeasily absorbed by the grid and they require less fromthe receiving country in terms of finance, construction,and oversight. Failing this, different deployment modelsmay help. For example, a fusion company that can offera BOO service (where it builds, fuels, operates, and de-commissions the plant) greatly reduces the burden onthe receiving country.
Process heat is used in a wide range of industrial pro-cesses, including chemical production, oil and steel re-fining, and paper production. It is a large market, com-prising 22% of primary energy use in the U.S. in 2014[21, p. 188]. However, it is a challenging market that willrequire special circumstances to be accessible to fusionas an early market. For example, a facility that faces un-usually high energy costs, such as a remote mine, couldbe an excellent candidate for early fusion deployment.One challenge is that process-heat plant designs varywidely, even within the same industry [22, p. 2], andmay require a custom retrofit in order to accept heatfrom fusion. Another challenge is that plant operatorsoften want to wait for several successful demonstra-tions in their industry before they consider wide-scaleadoption. We have heard from many experts that theprocess-heat industry is very conservative, eschewingrisks even when there is an opportunity to save money.Finally, consumers of industrial heat often require tem-peratures (e.g., > ◦ C) that may be challenging forfusion (or nuclear in general) to deliver (due to the tem-perature limitations of low-activation structural mate- otential Early Markets for Fusion Energy 7 rials in a neutron-irradiation environment), or they useeffectively free fuel that is a byproduct of their produc-tion process. Still, an industrial plant might choose tohave a colocated fusion power plant to supply electricityfor electrical process heating.The size of the process-heat market is best describedin [21, Appendix F], which presents a detailed analysisof the potential for fission to provide process heat, con-sidering the amount of power needed by each plant, thetemperature required, and whether the plant producesits own fuel as a byproduct of its operation. However,economics, siting restrictions, and less tangible issuessuch as public opinion are all ignored. Thus, the con-clusion that only 19% of the market, totaling 637 GW t worldwide [21, pp. 26–27], is a fit for fission heat ignoresmost of the characteristics that might give fusion an ad-vantage. In addition, the study makes two assumptionsthat may not be true for fusion. First, fission can pro-vide heat up to 850 ◦ C [21, p. 187]. In principle, fusionpower plants should have the same maximum tempera-ture range as fission plants (limited by materials prop-erties), but if the maximum temperature is higher orlower for fusion, the achievable part of the heat marketwill be larger or smaller, respectively. Second, the re-port assumes that fission power plants can be as smallas 150–300 MW t . Many fusion designs may need to belarger (dictated by physics) or combine several modulesof smaller size at a single site in order to amortize thecost of the tritium-processing plant. The minimum unitsize (with respect to power generation) of a single fu-sion core is not yet known and the subject of ongoingresearch. Hydrogen can be a substitute for fossil fuels for a widevariety of purposes, and can be used to make otherchemicals that can also substitute for fossil fuels. Forexample, hydrogen can be used instead of natural gasfor refining oil and iron and for heating buildings, andbe turned into ammonia for fertilizers and methanolfor synthetic gasoline. Today, hydrogen is almost exclu-sively made from fossil fuels. If zero-carbon hydrogen(“green hydrogen”) can be produced with fusion powerat competitive prices, then this could be a very largemarket for fusion. The price of hydrogen is more con-sistent around the world than the price of electricity,but still varies enough to allow for some promising po-tential early markets for fusion. Hydrogen productionuses a combination of heat and electricity but tends tobe produced in large facilities that can justify operatingtheir own power plant. As such, the market is differentenough to warrant exploring separately. In this section, we examine methods for producinghydrogen and plausible costs for producing it with fu-sion power, and then look at the more significant usesfor hydrogen. We discuss market sizes in megatons ofhydrogen (following the lead of the literature), whichcan be converted to a market for fusion energy by notingthat producing hydrogen with electrolysis uses about50 kWh e /kgH (or 5.6 GWyr/MtH ).5.1 Producing HydrogenHydrogen can be produced from hydrocarbons (usingthem both as a fuel and a source of hydrogen) or fromelectrolysis of water (using electricity to split water intohydrogen and oxygen). Without a carbon price, thelowest-cost way to produce hydrogen is from naturalgas or coal. With a high enough carbon price and im-provements in electrolysis and renewable energy, theprice ceiling would be set by electrolysis powered byrenewable energy.In summary, we find several potential early mar-kets for hydrogen produced with fusion power. Hydro-gen in China and Europe costs about 2.3 $ /kgH whenmade from natural gas with CCS [23, p. 42]. Greenhydrogen made with fusion-powered electrolysis couldbe competitive if the fusion power plants LCOE is ≤ $ /MWh e (for fusion-LCOE estimates, see “Electrol-ysis” subsubsection below). Without CCS, hydrogenfrom natural gas costs about 1.7 $ /kgH in these re-gions [23, p. 42], which would require a challenging fu-sion LCOE of 36 $ /MWh e . Western Europe and Japantogether could blend nearly 5 MtH /year into exist-ing natural-gas networks for heating without signifi-cant modification, with some fraction of this marketbecoming available if fusion costs 78 $ /MWh e (Japan)or 66 $ /MWh e (Europe). However, the above assumesan absence of large-scale competition from wind andsolar. Including the latter, fusion-powered electrolysismay require an LCOE of ≤ $ /MWh e to producehydrogen competitively. Details are discussed below. Today, almost all hydrogen is made from natural gas bypressurizing it and heating it with steam to break thehydrocarbons, or from coal by heating coal and waterto form “syngas,” and then separating out the CO [23,p. 42]. These two processes are called steam methanereforming and coal gasification, respectively. The costof these processes varies based on the local costs of gasand coal, but is in the range of 0.71–2.29 $ /kgH [24,p. 11]. Figure 5 shows the regional variation of theseprices. M. C. Handley, D. Slesinski, and S. C. Hsu
Fig. 5
Regional variation of hydrogen prices [23, p. 42].
Table 3
Possible improvements for cost of electrolysis [25];corresponding page numbers from [25] are given in the table.*See [25, Fig. 25], reading the midpoint between the conser-vative and optimistic costs for electrolyzers in China in 2035as 120 $ /kW.Characteristic Capital-cost ( $ /kgH peryear) reduction factorMW-scale instead of tens ofkW 90–97% (p. 23)Prices in China vs. theWest 50–80% (p. 23)Continued learning by 2035 35%* (p. 30)Integration with hydrogenplant 15% (by avoiding grid con-nections, p. 64) Using electricity to split water to produce hydrogen isenergy-intensive. However, with a low-enough cost ofenergy and high-enough capacity factor, hydrogen pro-duced by electrolysis can become competitive with hy-drogen produced from natural gas or coal. Using theassumptions described below, fusion LCOE would needto be 13–50 $ /MWh e to be competitive with fossil-fuelproduction of hydrogen.To determine this, we adjusted the model in [25] todescribe electrolysis powered by a fusion power plant.Because our calculations do not exactly match the pricesgiven in [25], as detailed in [26, Tab: Production, Row:Correct value for line above], all of the costs presentedhere are from our model, even for scenarios modeled in[25]. Electrolysis is presently very expensive (6–7 $ /kgH in North America and Europe [25, p. 28, Fig. 18]), butseveral significant improvements over the baseline aredescribed (see Table 3) and included in our model, asthey would all likely be implemented when integratinghydrogen and fusion plants. Figure 6 shows the low andhigh market costs for fossil-fuel production of hydrogen,corresponding to fusion LCOE of 13 and 50 $ /MWh e ,respectively.Electrolysis can also be powered by renewables, al-beit at a lower capacity factor. Many approaches are Fig. 6
Cost of producing hydrogen from fusion-powered elec-trolysis [26, Tab: Production, Column: Fusion, low-temp]. modeled in [25, p. 52], and the findings are summarizedin [25, p. 67, Figs. 62 and 63], including that a com-bination of wind and solar can deliver low-cost powerto an electrolyzer while still enabling a reasonable ca-pacity factor. We model the cost of this system in [26,Tab: Production, Column: China, optimized renewables2030] and estimate a cost of 1.52 $ /kgH . Fusion mustproduce power at ≤ $ /MWh e to compete with re-newable production of hydrogen in regions where windand solar power are plentiful. We cannot find detailed costs for high-temperature elec-trolysis, although the energy requirements for low- andhigh-temperature electrolysis are given in [27, p. 481].If the capital costs are the same for both, we find thatfusion LCOE would need to be 16–62 $ /MWh e for high-temperature electrolysis to compete with traditionalsources of hydrogen. This is only a modest improvementcompared to low-temperature electrolysis (requiring afusion LCOE of 13–50 $ /MWh e ) because a significantfraction of the energy is still supplied as electricity.Intermediate-temperature electrolysis (100–400 ◦ C)would get a larger fraction of its energy from electricitythan high-temperature electrolysis, and hence shouldbe less of a win unless the heat can be supplied fromwaste heat from the fusion power plant. We found evenfewer references for this, so we ignore intermediate-temperature electrolysis.There are yet other ways to produce hydrogen fromheat, including pyrolysis of natural gas, which uses heatto separate natural gas into hydrogen and carbon (notcarbon dioxide) [28], and thermo-chemical water split-ting, which uses heat and chemical reactions to sepa-rate water into hydrogen and oxygen [29]. Additionally,plasma catalysis uses electricity to produce hydrogen otential Early Markets for Fusion Energy 9 from methane [30]. Detailed cost analyses of these op-tions are beyond the scope of this paper.
In most hydrogen markets that we examine below, greenhydrogen does not compete with existing sources of hy-drogen so much as with existing markets/processes thatproduce and consume hydrogen internally. For example,the production of ammonia involves both endothermicand exothermic processes, and is more efficient as aresult of using heat from one to run the other. Thisproduces a lower implied price for hydrogen than themarket price. We examine the impact of a carbon priceon specific markets in the next subsection.Table 4 shows the cost impact of CCS on methanereformation and coal gasification, which emit significantamounts of CO and suffer reduced efficiency if pairedwith CCS. Table 5 shows the impact of carbon pricingbased on the numbers of Table 4. Without CCS, a priceof 50 $ /tCO would add 0.45 $ /kgH to the cost of hy-drogen based on methane reformation, and more than1 $ /kgH based on coal. At 50 $ /tCO , it is more eco-nomical for these plants to pay the carbon price ratherthan integrate CCS, whereas at 100 $ /tCO , the plantswould save money by integrating CCS. A high carbonprice can provide significant relief to green hydrogen, es-pecially when it is competing with hydrogen producedfrom coal.5.2 Uses of HydrogenIn this subsection, we examine some of the larger usesof hydrogen, not all of which are suitable early marketsfor fusion. As noted above, many of these uses of hydro-gen produce and consume hydrogen as part of a largerprocess, placing challenging cost constraints on greenhydrogen production. We ignore new uses for hydrogenon the basis that they will be more price-sensitive thanexisting uses. Hydrogen demand for ammonia production is predictedto be 38 MtH /year in 2030, i.e., 31 MtH /year in 2018[23, p. 99] and growing at 1.7% per year from 2018–2030 [23, p. 101]. While there are hopes for ammoniato play a major role as a fuel, today it is used almostexclusively in the chemical industry [23, pp. 61 and139], and is mostly produced from natural gas or coalin an integrated and exothermic process from hydro-gen and nitrogen. The energy released helps power the endothermic production of hydrogen by natural-gas ref-ormation or coal gasification. The result is an impliedprice of 0.35–2.08 $ /kgH when natural gas prices are inthe range 2–12 $ /MMBTU [31, p. 3], significantly lessthan the market price of green hydrogen. The highestprice we find for ammonia production is in China, cor-responding to 1.43 $ /kgH and a required fusion LCOEof 30 $ /MWh e . Table 6 shows the five highest costs fromthe top ten ammonia-producing countries.A carbon price provides significant breathing roomfor the price of hydrogen (summarized in Table 7). A100 $ /tCO price allows adding 0.9 and 1.5 $ /kgH when competing with natural gas and coal, respectively[31, p. 7]. We estimate that a 50 $ /tCO price wouldraise the competitive price for the bulk of the marketby about 0.5–1.17 $ /kgH . Unless the carbon pricingis carefully designed, it will not capture CO emissionsfrom the use of ammonia-based fertilizers, reducing itseffect [31, p. 7]. Hydrogen demand for methanol production in 2030 ispredicted to be 18 MtH /year, i.e., 12 MtH /year in2018 [23, p. 99] and growing at 3.6% per year from2018–2030 [23, p. 101]. Methanol (CH OH) is widelyused, e.g., in plastics and for blending into liquid fuels.However, the market is inaccessible to green hydrogenbecause the majority of methanol is produced in areaswith very low-cost natural gas and because of the costof acquiring CO feedstock that supplies the carbon inthe methanol.For example, producing methanol from green hydro-gen that costs 1 $ /kgH and carbon dioxide that costs95 $ /tCO cannot compete with methanol made fromthe most expensive natural gas or coal [32, p. 7]. Witha carbon price, the methanol facility can be paid forthe CO that it uses, instead of having to pay for it. At100 $ /tCO , the maximum carbon price that we con-sider, and with hydrogen at 2 $ /kgH , a methanol plantcan compete with the most expensive natural gas, buthydrogen under 1 $ /kgH is still required to competewith lower-cost natural gas. A 100 $ /tCO carbon pricemakes green methanol more competitive with methanolmade from coal: it only requires hydrogen between ap-proximately 2–3 $ /kgH to compete with low-cost andexpensive coal, respectively [32, p. 9]. Perhaps with a100 $ /tCO carbon price and low-cost hydrogen, thiscan be made to work, but the market seems difficultenough that we ignore it. Table 4
Impact of CCS on price of different hydrogen-production methods [24]; corresponding page numbers from [24] aregiven in the table.Production method Emissions Cost per 10 $ /tCO of Amortized cost of(kgCO /kgH ) associated emissions ( $ /kgH ) CCS ( $ /kgH )Methane reformation 8.9 (p. 4) 0.089 –Methane reformation w/CCS 0.89 (p. 11) 0.0089 0.6 (p. 11)Coal gasification 20.2 (p. 4) 0.202 –Coal gasification w/CCS 2.02 (p. 11) 0.0202 1.1 (p. 11) Table 5
Impact of various carbon prices on hydrogen prices,based on the information in Table 4.Carbon price Cost increase of Cost increase of( $ /tCO ) methane reformation coal gasification( $ /kgH ) ( $ /kgH )20 0.18 (no CCS) 0.40 (no CCS)50 0.45 (no CCS) 1.01 (no CCS)100 0.69 (w/CCS) 1.3 (w/CCS) Refineries currently use 38 MtH /year [23, p. 9], whichis expected to grow 7% by 2030 [23, p. 95]. About 20%of this is supplied by commercial sources [23, p. 95] atthe market prices described in the “Producing Hydro-gen” subsection above. It is possible that there are somerefineries that pay significantly more for hydrogen, butwe were unable to find data on this. The rest of thehydrogen consumed is produced onsite in already paid-for facilities or as a zero-cost byproduct of oil refining,both of which are difficult to undercut. We assume thatthere will not be significant construction of new oil re-fineries. A summary of the oil-refining market is shownin Table 8. Blast furnaces (BF) run on coal, do not need to be sup-plied with hydrogen, and produce 90% of the worldsiron [23, p. 108]. A less-common method of iron pro-duction is direct reduction (DR), which runs on natu-ral gas but can run on hydrogen instead. Coal and gasboth have wide price ranges but are primarily used inplaces where they have low cost, so the low end of theprice range is the relevant one.Without carbon pricing, green hydrogen can com-pete with DR powered by the lowest-cost natural gas(2 $ /MMBTU) at about 0.7 $ /kgH , opening up a 4-MtH /year market. Hydrogen must drop to 0.5 $ /kgH to compete with a blast furnace run on even moder-ately low-cost coal [33, p. 8]. Because this price is verylow and the blast furnaces would need to be replacedwith DR equipment to run on hydrogen, we ignored thislarger market in scenarios without carbon pricing. Carbon pricing significantly reduces the price pres-sure on hydrogen. A 100 and 50 $ /tCO price makes2 and 1 $ /kgH , respectively, competitive with all BFsteel and most DR steel, regardless of the fuel price. Ifall BF facilities converted to DR, then this would growthe market to 40 MtH /year. In the market summary(Table 9), we assume that all facilities would convertto DR and use green hydrogen in scenarios with a car-bon price. A carbon price of 20 $ /tCO would require0.7 $ /kgH to compete with the lowest-cost gas [33,p. 8]. Using pure hydrogen to heat buildings is challengingbecause it requires upgrades to all the equipment inthe gas network and coordination between many par-ties [23, p. 144]. However, various studies have foundthat hydrogen can be mixed with natural gas up to atleast 30% (by volume) without causing problems [23,p. 147], and markets appear able to support high prices.Table 11 shows the size of the hydrogen market for heat-ing buildings, assuming a 15% hydrogen mixture and aprice range for hydrogen to be competitive. Notably,the market starts to open up at 78 $ /MWh e in Japanand 66 $ /MWh e in Europe. We do not have data onthe size of the market in these countries at each pricepoint, but they should be large enough to support sev-eral early fusion plants [23, p. 149]. A carbon price adds0.09 $ /kgH [11, Tab: Heating] to the allowable hydro-gen prices for every 10 $ /tCO (Table 11). If environ-mental or economic pressures are large enough, thesemarkets might be able to double in size if they use 30%hydrogen instead of 15%. There is a large potential market, 370 Mtoe/year or 500GW, for using green hydrogen to produce high-gradeprocess heat, primarily for producing cement and var-ious chemicals [23, p. 116]. However, hydrogen is notcompetitive with other sources of heat even with a car-bon price of 100 $ /tCO [23, p. 118], and the marketis hard to address because of significant variation in otential Early Markets for Fusion Energy 11 Table 6
Highest prices of ammonia (NH ) from top producing countries [26, Tab: Ammonia].Interpolations Quantity of NH Levelized cost of Quantity of Required H Equivalent fusion(MtNH /yr) NH ( $ /kgNH ) H (MtH /yr) cost ( $ /kgH ) LCOE ( $ /MWh e )China (natural gas) 6 450 1.18 1.43 30Ukraine 5 430 0.94 1.33 27China (coal) 53 400 10.60 1.17 23India 13 370 2.60 1.01 20Indonesia 7 350 1.36 0.91 17 Table 7
Ammonia market summary.Carbon Fusion Hydrogen Marketprice LCOE price size( $ /tCO ) ( $ /MWh e ) ( $ /kgH ) (MtH /year)0 12 0.67 2720 17 0.87 2750 23 1.17 27100 35 1.67 27 Table 8
Oil refining market summary.Carbon Fusion Hydrogen Marketprice LCOE price size( $ /tCO ) ( $ /MWh e ) ( $ /kgH ) (MtH /year)0 20 1.00 820 24 1.18 850 30 1.45 8100 36 1.69 8 Table 9
Iron and steel refining market summary.Carbon Fusion Hydrogen Marketprice LCOE price size( $ /tCO ) ( $ /MWh e ) ( $ /kgH ) (MtH /year)0 8 0.5 420 13 0.7 450 20 1 40100 43 2 ∼ equipment even within an industry [22, p. 2]. Accord-ingly, we conclude that there is no early fusion marketfor this. There are several promising options for improving theeconomics of a fusion plant, including cogeneration us-ing the waste heat and by reducing the capital cost.6.1 CogenerationRevenue that can be extracted from waste heat aftergenerating electricity, i.e., cogeneration, can reduce theeffective LCOE and make a plant more competitive. This is true for any plant that produces heat, but fu-sion has some unique advantages in that it: (1) is zero-carbon, unlike fossil fuels, (2) should have more flexiblesiting options than a fission power plant, and (3) doesnot need favorable weather, unlike concentrated solarpower. This subsection explores the feasibility and prof-itability of cogenerating various products from fusionpower plants.
The increased need for desalination facilities in the fu-ture will create additional opportunities for fusion en-ergy. Assuming an LCOE of 50 $ /MWh e for the fusionpower plant, integrating a fusion and thermal desali-nation plant could lower the effective fusion LCOE byapproximately 30%, depending on the specific desalina-tion method.Due to over-exploitation of resources and effects ofclimate change, fresh-water supplies around the worldare declining. At the same time, population and eco-nomic growth will increase the demand for fresh water,with much of this growth in areas that already lack ac-cess to clean water. Global demand for all water uses,which is currently around 4600 km /year, is predictedto increase by 20–30% by 2050 to 5500–6000 km /year[34]. In addition, it is also projected that by 2050, 57%of the global population will live in areas that sufferwater scarcity at least one month each year [34]. Whilewater efficiency measures and further regulations couldhelp, additional sources of clean water may be needed.Desalination, which uses energy to remove salt fromwater to produce water suitable for drinking or agricul-ture, can be a solution to worsening water shortages.The first large-scale desalination plants were built in the1960s, and there are about 20,000 facilities in use today[35]. Because of the projected increase in demand forwater, it is likely that desalination will become a largerportion of the water market by the time fusion energyis commercially viable. If the predicted 20% demandincrease by 2050 is entirely supplied by desalination,this would create a need for approximately 700–1000GW (depending on the type of desalination used) [11, Table 10
Hydrogen market for heating buildings [26, Tab: Heating].Region Price, high Price, high Price, low Price, low Natural gas H demand at Electricity( $ /kgH ) ( $ /MWh e ) ( $ /kgH ) ( $ /MWh e ) demand for heating 15% of gas consumption to(Mtoe/year) (MtH /yr) produce (GW)Japan 3.5 78 2 43 14 0.73 4.19WesternEurope 3 66 2 43 80 4.19 23.92Korea 1.9 40 0.9 17 11 0.58 3.29Russia 1.8 38 1.5 31 43 2.25 12.86UnitedStates 1.5 31 1.2 24 147 7.70 43.95China 1.4 29 1.2 24 51 2.67 15.25Canada 1.2 24 0.8 15 21 1.10 6.28 Table 11
Heating buildings market summary.Carbon Fusion Hydrogen Marketprice LCOE price size( $ /tCO ) ( $ /MWh e ) ( $ /kgH ) (MtH /year)0 43 2 50 17 0.9 1920 20 1.0 1950 24 1.2 19100 31 1.5 19 Tab: Desalination Analysis] in order to power the newdesalination plants (energy needs are from [36, p. 40]).There are three main types of desalination tech-nology. These include two thermal processes, Multi-stage Flash Distillation (MSF) and Multi-effect Dis-tillation (MED), and one membrane process, SeawaterReverse Osmosis (SWRO). For MSF and MED desali-nation plants, heat and electricity are 52% and 14% ofthe O&M costs, respectively. For SWRO desalinationplants, electricity is 41% of O&M costs, with no heatrequired [36, p. 27]. MSF technology can produce waterat scale at lower cost than MED technology, which ismore common for smaller thermal desalination plants[36, p. 21]. SWRO also can offer significant economiesof scale at smaller sizes, but those taper off above plantcapacities of 100 MLD (million liters per day). However,with no heat requirements, SWRO alone will likely notbe the best fit for fusion. Some desalination plants arealso built as hybrid plants, in order to incorporate acombination of either MSF or MED with SWRO, thusallowing for the plant to take advantage of the lowest-cost available thermal or electrical energy, while stillallowing 24-hour water production.We estimate the savings from powering a desalina-tion plant using waste heat from a fusion power plant.Both thermal desalination methods require low-gradeheat: around 100 ◦ C and 70 ◦ C for MSF and MED, re-spectively [37, p. 12]. Fusion can take advantage of theserequirements by supplying waste heat at little-to-no cost. The cost components of an MSF and MED plantthat can deliver water at 1.07 $ /m and 0.83 $ /m ,respectively, are described in [36, p. 47], assuming anLCOE of 50 $ /MWh e . We assume that the fusion powerplant, which is sized to supply all its waste heat to meetthe desalination requirements, provides heat and elec-tricity to the desalination plant, and that each MWh t used by the desalination plant reduces the electricaloutput of the fusion power plant by 0.15 MWh e [38,p. 50]. We then estimate the revenue that the fusionpower plant receives from desalinated water as the dif-ference between the cost of running the desalinationplant with grid connections and the cost of running itwith heat and electricity from the fusion power plant.Incorporating these factors, we find that cogenerationusing MSF and MED lowers the effective fusion LCOE27% (from 50 to 37 $ /MWh e ) and 35% (to 32 $ /MWh e ),respectively [11, Tab: Desalination Analysis].While this analysis may make MED appear to bethe best desalination type to combine with fusion, inreality it depends on the size of the desalination facil-ity. Although MSF has slightly higher capital costs andenergy requirements than MED, it is easier to operateand benefits more from economies of scale at highercapacities, such as 500 MLD or greater [36, p. 45]. Al-though this would be considered a mega-size desalina-tion project, a plant would likely need to be in this sizerange in order to economically pair with fusion. For ex-ample, based on average energy requirements for MSF,a 225-MW e fusion plant would be able to deliver all itsenergy to a 500-MLD desalination plant [11, Tab: De-salination Analysis]. Using fusion energy to power de-salination could provide a unique opportunity to openadditional markets for fusion, reduce the effective costsof a fusion plant, and produce desalinated water in aclean and sustainable way. otential Early Markets for Fusion Energy 13 Powering a direct-air-capture (DAC) plant with wasteheat from a fusion power plant could reduce the effec-tive fusion LCOE. Assuming contemporary costs of aDAC plant and an LCOE of 50 $ /MWh e for the fusionpower plant, integrating the two plants could lower theeffective fusion LCOE by about 35%.According to the IPCC, any chance to keep theglobal average temperature increase from pre-industriallevels under 1 . ◦ C will require some form of CO re-moval [39]. One solution for removing CO is DAC,which pulls CO directly from the atmosphere at amuch lower concentration than point-source carbon cap-ture. DAC technology is relatively new, energy-intensive,and expensive. However, pilot plants have been built,and new technologies with increased scale will reducecosts. It is also likely that by the time fusion energy be-comes commercially viable, the demand for DAC maybe much higher. A conservative estimate for negativeemissions eventually needed to reach global climate tar-gets is about 10 GtCO per year [40]. If this amountwere to come entirely from the solid-sorbent DAC method(see below), it would require roughly 1300 GW of heatand 270 GW of electricity. This increase in energy de-mand could become a significant market for fusion en-ergy [11, Tab: DAC Analysis].Currently, there are two main types of DAC tech-nology: a high-temperature, liquid-solvent method, anda low-temperature, solid-sorbent method. Both requiremostly heat, as well as some electricity. The liquid-solvent and solid-sorbent methods require heat at ap-proximately 900 ◦ C and 100 ◦ C, respectively [41, p. 962].Because the upper temperature limit for early fusionpower plants is not yet known, we focus on the solid-sorbent method in our analysis.Consider the example of a solid-sorbent DAC plantdescribed in [41, p. 965]. The plant captures 360,000tons of CO per year at 187 $ /tCO , assuming an LCOEof 50 $ /MWh e and a levelized cost of heat (LCOH) of20 $ /MWh t . We assume that the fusion power plant,which is sized to supply all its waste heat to meet theDAC requirements, provides heat and electricity to theDAC plant, and that each MWh t used by the DACplant reduces the electrical output of the fusion powerplant by 0.15 MWh e [38, p. 50]. We then estimate therevenue that the fusion power plant receives from cap-tured CO as the difference between the cost of runningthe DAC plant with grid electricity and the cost of run-ning it from heat and electricity from the fusion powerplant. (In this analysis, the price of DAC when payingfor energy was more than our 100- $ /tCO threshold fora carbon price. In the long run, we expect prices to be lower, but we used the best data that we had for thisanalysis.) Incorporating all these factors, we found thateven with a reduction in electricity sold, the revenuefrom the DAC plant lowers the effective fusion LCOEfrom 50 to 32 $ /MWh e , i.e., a 35% decrease [11, Tab:DAC Analysis]. Although this analysis has uncertainty,it suggests that using a fusion power plant to powera DAC plant could reduce the effective fusion LCOE,assuming that there is a market for captured CO .Lastly, when compared to desalination, DAC as aneconomic boost for fusion has somewhat lower near-term potential. This is because both analyses dependon the assumption that the plants take in revenue equalto the cost of capturing carbon or producing desali-nated water. Athough the situation could easily changeover the next few decades, there is presently very littlemarket demand or financial incentive to pay for CO ,whereas there is already a well-established and increas-ing demand for fresh water in many areas. Powering a district-heating system with fusion appearsto be a promising long-term market, but in the shortterm it requires a large, pre-existing heat network, whichseems to be a rare opportunity.District heating provides heat to buildings from ashared heat source, which produces steam or hot waterthat is piped to each building and heat-exchanged withthe buildings heating system. This can be an econom-ical alternative to on-site combustion of fossil fuels orheating via electricity (an expensive form of energy),especially when using heat sources that do not scaledown to the power required by a single building, e.g.,fission or fusion power plants. In the long term, it al-most certainly makes sense for fusion to power district-heating networks. The world uses 17.5 PWh t of low-grade ( < ◦ C) heat annually [42, p. 48]. Individualcities in northern Europe could each use multiple GW t of district heating [38, p. 58]. When provided by a fis-sion or fusion power plant, each MWh t extracted onlyreduces electricity production by 0.15 MWhe [38, p. 50].If fusion power plants can be sited close to populationcenters, the pipes between the heat source and the heatmarket could be shortened, saving hundreds of millionsto billions of dollars [38, p. 55].However, it is unclear that district heating makessense as an early market for fusion. Existing heat net-works are small: 17,000 networks in the UK togethercarry 1.4 GW in total [38, p. 48], and where they areheated by fission power plants, they use an averageof 5% of the plants heat [42, p. 48]. We find it un-likely that using 5% of the waste heat of a fusion power plant for heating will change the economics of the plant.However, a coal power plant with a significant district-heating load on its waste heat would be an excellentcandidate for retrofitting with a fusion power plant,capturing extra revenue from heating as well as costsavings from reusing part of the coal power plant, asdetailed in the “Retrofitting Coal Plants” section be-low.To be suitable for district heating, a fusion powerplant needs to be able to supply hot water or steam at < ◦ C. Modifying existing fission power plants to dothis is relatively straightforward [42, p. 48], and thusshould probably not be a problem for fusion plants ei-ther. Given that this market remains small for now, werecommend that fusion developers focus on this only ifit proves necessary to make the economics work and/orfind a district-heating opportunity that uses a largefraction of the waste heat.6.2 Lowering Capital CostsIn order to fully compete with NGCC power plants, fu-sion companies should look to minimize the overall cap-ital that must be risked to build a fusion power plant.Fuel is 80% of the LCOE of an NGCC power plant,which makes the construction of a new plant relativelylow financial risk. By contrast, capital costs are 80% ofthe LCOE from fission [21, p. 34] and will likely be sim-ilar for fusion. Even if two power plants, e.g., one fusionand one NGCC, have the same LCOE, the fusion powerplant will be a much larger financial risk because moreof the costs committed during construction will be lostif the plant is never completed or becomes unprofitableto operate. This fact is partly hidden by the metricsused by the industry.
When considering the cost to construct a plant, it is im-portant to use the correct metrics. Often, power-plantdevelopers quote the overnight construction (or capital)cost (OCC), which is the cost in dollars per watt ( $ /W)required to build a plant if it could be constructed in-stantaneously, i.e., “overnight.” The first problem withOCC is that it ignores construction time and the costof capital, which can be substantial. For example [43,p. 17], a fission power plant that takes 10 years to buildwith an 8% interest rate (a high-risk project) will cost70% more than an NGCC power plant that takes twoyears to build and has a 6% interest rate, even thoughthey have the same OCC [11, Tab: OCC/TCC]. Thesecond problem is that OCC does not measure finan-cial and bankruptcy risks. The latter are better mea- sured by the total construction (or capital) cost (TCC),which drives decisions about whether plants get built.TCC, measured in dollars rather than $ /W, is the sumof all direct and indirect (including interest) costs un-til the plant is producing revenue. Fusion developersshould avoid the pitfall of designing a very large plant toachieve low $ /W. Determining a range of recommendedfusion TCCs is beyond the scope of this report, but werecommend adopting methods to minimize constructiontime and TCC, e.g., see [21, Sec. 2.5]. Another potential way to reduce TCC is to reuse equip-ment in a coal power plant. A fusion core that candeliver steam at 500–600 ◦ C can save up to 30% onTCC by retrofitting an existing coal plant. Because fu-sion LCOE is likely to be dominated by capital costs,retrofitting a coal plant could also significantly reduceLCOE.A recent study [44] examined retrofitting an exist-ing coal power plant to generate clean power throughvarious combinations of solar, wind, enhanced geother-mal systems, biomass, CCS, or small modular fissionreactors. The study found that fission reactors wouldbe the most viable retrofit option for coal power plantswhile generating a similar amount of electricity. How-ever, there are some caveats to this finding, as it wouldnot be useful to retrofit coal plants of any age or size.The most applicable coal plants are ones with greaterthan 50-MW capacity and either recently built or mod-ernized, meaning most of the valuable equipment on thesite is less than 20 years old [44, p. 12]. Although likelyreduced by the time fusion is commercially viable, thereare currently about 1,100 coal plants worldwide with acombined capacity of roughly 1.1 TW that fall into thiscategory [45]. It is worth noting that about 800 of thesecoal plants are in China. In addition, some have capaci-ties in the range of multiple GW, which could allow formultiple fusion cores to be built on the same site.To maximize the cost savings of installing a fusionenergy system at a coal site, all applicable equipmentshould be reused. Ordered from easiest to most difficult,this includes the site location, grid connection, electricalequipment, external heat interfaces, turbine and gener-ator, and steam-cycle cooling equipment. In order toreuse the full steam cycle, the thermal heat output ofthe fusion plant would need to generally match, or bea multiple of, the thermal output of the coal unit [44,p. 17]. It should be noted that even with reusing ev-erything up to the full steam cycle, only 40% of thecoal-plant costs will be recovered, as the other 60% is otential Early Markets for Fusion Energy 15 dominated by the coal, ash, and flue-gas systems thatare not needed [44, p. 7].By comparing the savings anticipated when repow-ering a 460-MW t coal power plant with an advancedfission reactor [44, p. 27] against cost estimates froman update [46] to a study [47, pp. 25–26] for a 150-MW e fusion power plant (corresponding to 460 MW t assuming 33% conversion efficiency), we find that a fu-sion retrofit might save 22–40% of capital costs acrossfour fusion plant designs [11, Tab: NEW repoweringcoal plants]. Furthermore, significant job retention maybe possible compared to abandoning coal plants as astranded asset [44, p. 33].While there are uncertainties needing more carefulexamination, it appears that TCC can be meaningfullyreduced by retrofitting a coal power plant with fusion. In this paper, we surveyed many energy sectors to iden-tify the most promising potential early markets for fu-sion energy in the 2035 time frame, taking into ac-count various scenarios including “business as usual,”high penetration of renewables, and carbon pricing. Wetook a relatively unforgiving view (in light of antici-pated continued cost reductions of the competition suchas renewables) in order to provide conservative recom-mendations, resulting in generally aggressive cost tar-gets that fusion must hit to break into its first markets.Our key, high-level findings are summarized in the sec-ond section of the paper. However, we reiterate thatif and when fusion is proven at scale with reasonablecosts (even if initially higher than the competition),it is likely to be a disruptive energy technology thatmay fundamentally and eventually alter markets andthe way humans use energy. Fusion may well deliver onthe promise of clean, abundant, safe, low-cost energy,allowing it to replace many other energy sources. Itslong-term market is potentially enormous.Finally, although this paper focused on commercialenergy-related markets, fusion may also have a rangeof space applications, e.g., see [48,49], with differenttechno-economic requirements. Assessing space appli-cations as a potential early market for fusion, thoughwarranted, was beyond the scope of this paper.
Acknowledgements
We are grateful for the insights and in-put provided by many people, especially Bob Mumgaard,Brandon Sorbom, Shiaoching Tse, and Ally Yost (Common-wealth Fusion Systems), Joe Chaisson (Clean Air Task Force,Energy Options Network), Eric Ingersoll (Lucid Catalyst),Armond Cohen (Clean Air Task Force), and Richard Pearson(Kyoto Fusioneering). Reference herein to any specific non-federal person or commercial entity, product, process, or ser- vice by trade name, trademark, manufacturer, or otherwise,does not necessarily constitute or imply its endorsement, rec-ommendation, or favoring by the U.S. Government or anyagency thereof or its contractors or subcontractors.
References
1. C.L. Nehl, R. Umstattd, W.R. Regan, S.C. Hsu, P.B.McGrath, J. Fusion Energy , 506 (2019)2. ARPA-E ALPHA program, https://arpa-e.energy.gov/technologies/programs/alpha
3. ARPA-E BETHE program, https://arpa-e.energy.gov/technologies/programs/bethe
4. ARPA-E GAMOW program, https://arpa-e.energy.gov/technologies/programs/gamow
5. T.L. Peckinpaugh, R.P. Stimers, M.L. O’Neill (2018).Invigorated Federal Interest in Fusion Energy PresentsOpportunities and Questions for Growing PrivateFusion Energy Sector,
6. A.C. Roma, S.S. Desai (2020). The Regulationof Fusion–A Practical and Innovation-Friendly Ap-proach,
7. Fusion Industry Association (2020). Ig-niting the Fusion Revolution in America,
8. D.R. Lewis, J.S. Merrifield, S.L. Fowler (2020).Considerations for Regulation of Fusion-Based Power Generation Devices,
9. DOE, NRC, and FIA Public Forum (2020). VirtualPublic Forum on a Regulatory Framework for Fusion, https://science.osti.gov/fes/Community-Resources/Workshop-Reports/DOE-NRC-FIA-Public-Forum
10. World Bank Group,
State and Trends of Carbon Pricing2019 (World Bank, Washington, DC, 2019). http://hdl.handle.net/10986/31755
11. See supplemental material: Electricity spreadsheet, avail-able at http://bit.ly/3pD43Ti (update with journal-provided URL upon publication)12. E. Ingersoll, K. Gogan, J. Herter, A. Foss (2020). Costand Performance Requirements for Flexible AdvancedNuclear Plants in Future U.S. Power Markets,
13. C. Turchi, G. Heath (2013). Molten Salt Power TowerCost Model for the System Advisor Model,
14. Lazard (2019). Lazards LevelizedCost of Energy Analysis–Version 13.0,
15. U.S. Energy Information Administration (2020).Levelized Cost and Levelized Avoided Cost of New Gen-eration Resources in the Annual Energy Outlook 2020,
16. J. Platt, J.O. Pritchard, D. Bryant (2017). Analyzingenergy technologies and policies using DOSCOE, https://dx.doi.org/10.2139/ssrn.3015424
18. H. Ritchie, M. Roser (2017). Fossil Fuels, https://ourworldindata.org/fossil-fuels
19. N.A. Sepulveda, J.D. Jenkins, F.J. de Sisternes, R.K.Lester, Joule , 2403 (2018). https://doi.org/10.1016/j.joule.2018.08.006
20. M. Ford, A. Abdullah (2020). New Methods For Evaluat-ing Energy Infrastructure Development Risks, submittedfor publication21. MIT Energy Initiative (2018). The Future ofNuclear Energy in a Carbon-Constrained World, https://energy.mit.edu/wp-content/uploads/2018/09/The-Future-of-Nuclear-Energy-in-a-Carbon-Constrained-World.pdf
22. BloombergNEF (2019). Hydrogen: The Economics of In-dustrial Heat in Cement, Aluminum and Glass, https://about.bnef.com/product
23. IEA (2019). The Future of Hydrogen,
24. BloombergNEF (2020). Hydrogen: The Economics ofProduction from Fossil Fuels With CCS, https://about.bnef.com/product
25. BloombergNEF (2019). Hydrogen: The Economics ofProduction from Renewables, https://about.bnef.com/product
26. See supplemental material: Hydrogen spreadsheet, avail-able at http://bit.ly/3pD43Ti (update with journal-provided URL upon publication)27. S.P.S. Badwal, S. Giddey, C. Munnings, Energy and En-vironment , 473 (2012). https://doi.org/10.1002/wene.50
28. M. Serban, M.A. Lewis, C.L. Marshall, R.D. Doctor,Energy Fuels , 705 (2003). https://doi.org/10.1021/ef020271q
29. Hydrogen Production: Thermochemical WaterSplitting,
30. M.L. Carreon, Plasma Res. Express , 043001 (2019). https://doi.org/10.1088/2516-1067/ab5a30
31. BloombergNEF (2019). Hydrogen: Making Green Am-monia and Fertilizers, https://about.bnef.com/product
32. BloombergNEF (2019). Hydrogen: The Economics ofLow-Carbon Methanol, https://about.bnef.com/product
33. BloombergNEF (2019). Hydrogen: Making Fossil-FreeSteel, https://about.bnef.com/product
34. A. Boretti, L. Rosa, npj Clean Water , 15 (2019). https://doi.org/10.1038/s41545-019-0039-9
35. J. Robbins (2019). As Water ScarcityIncreases, Desalination Plants Are onthe Rise, https://e360.yale.edu/features/as-water-scarcity-increases-desalination-plants-are-on-the-rise
36. World Bank Group,
The Role of Desalination in an In-creasingly Water-Scarce World (World Bank, Washington,DC, 2019). http://hdl.handle.net/10986/31416
37. IRENA (2012). Water Desalination Us-ing Renewable Energy, https://irena.org/-/media/Files/IRENA/Agency/Publication/2012/IRENA-ETSAP-Tech-Brief-I12-Water-Desalination.pdf
38. S. Qvist (2020). Nuclear Power Plants as Integrated En-ergy Centres, unpublished39. IPCC (2018). Global Warming of 1 . ◦ , https://report.ipcc.ch/sr15/pdf/sr15_spm_final.pdf
40. G. Peters, O. Geden, Nature Clim. Change , 619 (2017). https://doi.org/10.1038/nclimate3369
41. M. Fasihi, O. Efimova, C. Breyer, J. Clean. Prod. ,957 (2019). https://doi.org/10.1016/j.jclepro.2019.03.086
42. IAEA (2017). Opportunities for Cogeneration with Nu-clear Energy, Report No. NP-T-4.1,
43. Clean Air Task Force (2018). Advanced Nuclear En-ergy,
44. S. Qvist, P. Gladysz, L. Bartela, A. Sowizdzal, Energies , 120 (2021). https://doi.org/10.3390/en14010120
45. World Resources Institute (2019). Global PowerPlant Database, https://datasets.wri.org/dataset/globalpowerplantdatabase.
46. S. Woodruff, E. Ingersoll, A. Foss, R.L.Miller (2020). Revisit of the 2017 Costingfor Four ARPA-E ALPHA Concepts, https://arpa-e.energy.gov/technologies/publications/updated-cost-study-and-final-report-four-alpha-fusion-concepts
47. S. Woodruff, R. Miller, D. Chan, S. Routh, S. Basu,S. Rao (2017). Conceptual Cost Study for a Fusion PowerPlant Based on Four Technologies from the DOE ARPA-E ALPHA Program, http://dx.doi.org/10.13140/RG.2.2.24116.55688
48. J. Cassibry, R. Cortez, M. Stanic, A. Watts, W. Seidler,R. Adams, G. Statham, L. Fabisinski, J. Spacecraft Rock-ets , 595 (2015). https://doi.org/10.2514/1.A32782
49. G.A. Wurden, T.E. Weber, P.J. Turchi, P.B. Parks, T.E.Evans, S.A. Cohen, J.T. Cassibry, E.M. Campbell, J. Fu-sion Energy , 123 (2016)., 123 (2016).