Not Dead Yet

Global Nuclear Industry Picked Up Steam in 2015


A Gen-II nuclear plant under construction in Hainan, China.

April 22, 2016 | Will Boisvert,

Last year the success of wind and solar power made headlines as installations of new turbines and PV panels soared. Meanwhile, “nuclear is dead” think pieces mushroomed in the press as old plants closed and new projects floundered in delays and cost over-runs.

But while the “rise of renewables” is indeed reason to celebrate, the “death of nuclear” storyline has been greatly exaggerated. Far from being moribund, in 2015 the global nuclear sector quietly had its best year in decades. New reactors came on line that will generate as much low-carbon electricity as last year’s crops of new wind turbines or solar panels. The cost of building those reactors was less than one third the cost of building the wind turbines and solar panels, and typical construction times were under 6 years. The conventional wisdom that nuclear projects must be decade-long, budget-busting melodramas proved starkly wrong last year. In crucial respects the nuclear renaissance has hit its stride and is making a fundamental contribution to decarbonization—one that will accelerate if the industry gets recognition and support for what it is doing right.

Last year ten reactors—eight in China, one in South Korea, and an experimental fast reactor in Russia—connected to the grid with 9.4 gigawatts (GW) total capacity, twice as much new capacity as in 2014 and the most since 1990.1 That pace should accelerate in 2016, with at least eleven more reactors expected to come on line in China, South Korea, Russia, India and the United States.2

How does that stack up against wind and solar? At first glance it seems like last year’s worldwide addition of 64 GW of wind power3 and 59 GW of solar power4 dwarfed the new nuclear capacity. But that raw count can be misleading, because a wind or solar gigawatt is not the equal of a nuclear gigawatt in productivity and longevity. Wind turbines have an average capacity factor—the electricity they generate divided by what they could generate if they ran at full capacity all the time—of about 25 percent,5 while solar power clocks in at about 15 percent on average.6 Those low capacity factors reflect the productivity deficit from the fickleness of wind and sun. Nuclear reactors are not weather-dependent and can produce at maximum power most of the time, so they have much higher capacity factors: close to 90 percent for the Chinese and South Korean pressurized water reactors, and about 75 percent for the Russian reactor.7 Nuclear plants also last longer. Most reactors in the US have received license extensions to 60 years, and applications for 80-year extensions are in the offing; wind farms last only 20 to 30 years while solar plants last 25 to 40 years. With a higher capacity factor and longer service life, a nuclear gigawatt can generate six to eight times as much electricity over its lifespan as a wind or solar gigawatt.

Last year’s crop of nuclear reactors therefore has a productive potential comparable to that of the new wind or solar capacity. Those 9.4 GW of nuclear power will generate 71 terawatt-hours (trillion watt-hours, TWh) each year, close to the 78 TWh produced by 59 GW of solar power. The 64 GW of wind capacity will generate twice as much electricity per year, 140 TWh, but over a 60-year lifespan the reactors will produce the same amount, about 4,200 TWh, as the wind turbines will produce during a 30-year service life—and 35 percent more than the 3,120 TWh the solar capacity will produce in 40 years.  Last year’s new reactors will thus make as large a total contribution to the world’s low-carbon energy supply as the new wind turbines, and substantially more than the new solar panels.


New Global Wind, Solar and Nuclear Capacity in 2015

Assumes capacity factors of 25 percent for wind, 15 percent for solar and 85 percent for nuclear
Cost assumptions in footnotes 10, 11 and 12

And the nuclear capacity cost much less. Wind and solar generators have seen sensational drops in construction costs to as low as $1,300 to $1,400 per kilowatt for large-scale installations in some places. But because of its greater productivity, last year’s new nuclear capacity was cheaper. “Overnight” construction costs, excluding financing costs, for reactors in both China and South Korea ran about $2,300 per kilowatt; adding financing costs brings that to about $3,100 per kilowatt.8 (The Russian reactor cost about $4.8 billion for 789 megawatts of capacity.)9 So while last year’s increment of new wind capacity cost about $109 billion10 and new solar over $92 billion,11 the ten new reactors cost about $31 billion12—one third the cost or less, for nuclear plants that can equal or outstrip the life-cycle electricity production from last year’s wind or solar additions. A dollar invested in new nuclear reactors thus yielded, on average, more clean energy than one invested in wind or solar power.

That comparison is especially telling in China, which is deploying both renewables and nuclear on a massive scale.

New Chinese Wind, Solar and Nuclear Capacity in 201513

Assumed costs of $1,300 per kilowatt for wind and solar, $3,100 per kW for nuclear.
Assumed capacity factors of 22 percent for wind, 15 percent for solar and 86 percent for nuclear.

These numbers show that the huge preponderance of nominal wind and solar gigawatts in China is outweighed by the greater output and lifespan of nuclear gigawatts. In 2015 the reactors China completed cost a little over one third as much as the new wind and solar generators, but can produce over twice as much electricity each year as the solar panels and almost as much as the wind turbines. Over a 60-year service life the reactors will produce 20 percent more electricity than the new wind and solar capacity combined over a shorter 30- to 40-year lifespan. Despite adding four times more wind capacity than nuclear capacity, China saw its nuclear generation grow more than its wind generation in 2015, 38 terawatt-hours for nuclear compared to 32 terawatt-hours for wind. The country got almost as much electricity from its nuclear sector as a whole as it did from its wind sector, 171 TWh to 185 TWh,14 even though the total grid-connected wind capacity, 128 GW, was almost five times larger than the country’s 27 GW of nuclear capacity.

Those disparities in productivity yield a cost advantage for nuclear power that is reflected in Chinese electricity prices: new nuclear plants received a feed-in tariff of 0.43 yuan (6.5 cents) per kilowatt-hour in 2015,15 while wind capacity got 0.49 to 0.60 yuan and solar 0.9 to 1.0 yuan.16 China isn’t an anomaly; in South Korea, nuclear power is the cheapest electricity on the grid—cheaper even than coal-fired power.17 These per kilowatt-hour price comparisons do not count the additional grid costs of intermittent renewables, like the billions of dollars being spent on transmission lines to relieve curtailments of wind and solar surges, which wasted 15 percent of China’s wind output and 9 percent of its solar output last year.18 Nor do they capture the higher quality and reliability of nuclear electricity.

Nuclear Costs, High and Low

Last year’s experience shows that nuclear can often be the most economical way to bring clean electricity onto the grid. That’s nothing new. Although the nuclear industry has a reputation for ever-escalating costs, budget blowouts have been more the exception than the rule outside of the United States. A peer-reviewed study by the Breakthrough Institute’s Jessica Lovering, Arthur Yip and Ted Nordhaus shows that in France, Germany, Canada, Japan and India nuclear construction costs have been fairly stable—and low.19

China and South Korea are following that pattern by playing from a familiar playbook: leveraging economies of series and expertise through large-scale, systematic deployments of mature reactor designs with long production runs. Seven of last year’s new reactors were the Chinese CPR-1000 model, an updated version of the workhorse Generation II light-water designs that make up most of the world’s reactor fleet.20 China has now built 14 units of this design. South Korea’s reactor was the tenth unit of its OPR-1000 model, another updated Gen II design.21 In both countries, the nuclear industry is a partnership between state-owned utilities and construction companies with decades of continuous experience. With proficient project managers and construction crews and well-developed supply chains, the average construction time for these reactors was 5 years and 8 months.22 (The Korean project was delayed a year to yank out and replace electric cables that were discovered to have fake safety certifications.) The epic length and cost of nuclear projects in the West aren’t typical of the global industry.

These successes illuminate what’s gone wrong in the United States and Europe. The great model change-over to Generation III+ reactors with novel design features has gotten mired in delays and over-runs in Western countries that haven’t built new reactors for decades. The EPR reactor, designed by the French company Areva, is the poster-child of nuclear dysfunction: construction of the Flamanville unit will take eleven years and its price has tripled to 10.5 billion euros ($7,200 per kilowatt).23 The EPRs proposed for the Hinkley C nuclear plant in Britain are budgeted at a whopping GBP 24.5 billion, or $10,542 per kilowatt (financing included).24 Many explanations for these huge costs and delays, both plausible (high wages and regulatory red tape) and far-fetched (a “negative learning curve”), have been proposed. But the deeper problems are basic stumbling blocks to industrial efficiency: immature reactor designs that haven’t had the bugs worked out; inexperienced managers, workers and suppliers; and sporadic, piece-meal deployments that don’t let builders develop expertise.

Not all Gen III projects are fiascos. South Korea grid-connected its first APR-1400 reactor, a Gen III model, in January after seven years of construction (including a two-year delay to replace faulty cables and valves); overnight construction costs for the reactor are $2,450 per kilowatt.25 Japan built four Gen III ABWR reactors in the 1990s and 2000s, on schedule and for an overnight cost averaging $3,000 per kilowatt.26 These projects were successful in part because the models are incremental developments of familiar designs, and because the South Korean and Japanese nuclear construction industries had not rusted during a decades-long hiatus, as happened in the West.

The flagship Western Gen III+ designs, Areva’s EPR and Westinghouse’s AP-1000, are a different story. These designs feature more radical departures from previous models. The EPR is a study in complex redundancy that includes two nested containment walls. The AP-1000 has an innovative passive safety system that uses gravity, convection and cycles of evaporation and condensation to cool the reactor in an emergency. These models were supposed to be cheaper and faster to build than Gen II designs. Unfortunately, all the EPR and AP-1000 projects, in the West and China, have struggled with setbacks and overruns.

The Vogtle project in Georgia, where two AP-1000s are being built beside the plant’s two existing reactors, shows how excruciating the teething pains can be for a new model. The plant’s novel design has caused innumerable delays, gloomily chronicled in oversight reports from the Georgia Public Service Commission’s staff.27 They began even before construction officially started, when the AP-1000’s license approval was delayed for months while the Nuclear Regulatory Commission vetted the redesign of the shield wall for the containment building that houses the reactor.

Indeed, the AP-1000 design was so new that in important respects it was not even finished when they started building it. The “engineering packages”—the detailed drawings and specs needed to build the plant’s parts and structures—hadn’t been completed by Westinghouse when construction began in 2012, and delays in furnishing them continue to impede contractors [39]. When they were available they sometimes proved faulty. Flaws in the specs led to a seven-month delay in the pouring of one reactor’s concrete base mat foundation while improperly installed rebar was torn out and replaced. Some of the sub-modules—the pre-fabricated assemblies that comprise much of the plant—had to be redesigned because the original plans proved “impossible to physically construct,” according to GPSC staff.28 New NRC licensing procedures were expected to eliminate difficulties encountered on previous builds when mid-construction design changes imposed by regulators forced lengthy delays; unfortunately, the immaturity of the AP-1000 design itself has brought similar problems to Vogtle and other projects.

Design missteps have been matched by pitfalls in manufacturing and construction. A centerpiece of the AP-1000 design is its modular construction philosophy: factory-built submodules would be shipped to the site, assembled into modules and hoisted into place. This method is supposed to reduce cost and construction time by replacing much on-site construction with more efficient factory construction; to implement it, a sub-module factory was built in Lake Charles, Louisiana. But practice fell woefully short of theory. Workmanship at Lake Charles was poor, inspection and documentation of quality standards sloppy (the NRC doesn’t take kindly to that), and time was wasted when defective submodules got sent back for re-work. Production lagged way behind due dates. [40] The problems grew so bad that Westinghouse outsourced some of the submodule fabrication to other contractors, who in turn struggled with slow pace and quality issues. Fabrication of the shield wall’s panels, a novel design consisting of a sandwich of steel plates with concrete filling, also fell behind schedule. Even basic on-site construction tasks at Vogtle—welding and concrete-pouring, which is what people mainly do when they build a nuclear plant—have been plagued by poor quality and re-work, shortages of skilled tradesmen and low productivity. America’s long-dormant nuclear construction industry—the last plant went on line in 1996—wasn’t yet fully awake for the project.

With all this muddle, the construction time for the Vogtle reactors has swollen to almost eight years and the budget has risen 29 percent to $17.2 billion,29 putting it up in EPR nose-bleed territory at $7700 per kilowatt (financing included). More delays may follow. As the GPSC staff puts it, that’s what happens when “the design is new, the modular construction of nuclear power plants is new, the regulatory environment…is new and most of the people involved are new to new nuclear construction.”30

But the travails of Vogtle and other builds shouldn’t toll the bell for nuclear power. For one thing, the project’s huge expenses still aren’t all that expensive. GPSC staff estimate the two new units’ lifetime “revenue requirement”—the money needed to recoup all the expenses, both capital costs and annual operating and maintenance costs over the plant’s service life—at $65 billion.31 That sounds like a lot, but spread over their 60-year output, a prodigious 1,056 terawatt-hours, it comes to 6.2 cents per kilowatt-hour. That’s not the cheapest power around but it’s not outrageously expensive. (Vogtle’s parent utility Georgia Power paid 4.33 cents per kilowatt-hour to buy wholesale power last year.)41 South Carolina’s V. C. Summer project, with another two AP-1000s, is coming in considerably cheaper at $5,800 per kilowatt (financing included),32 while recent estimates for the first Chinese AP-1000 plant at Sanmen put overnight costs at about $2,700 per kilowatt.42

And it is important to acknowledge Vogtle’s challenges as what the industry politely calls “first-of-a-kind” issues. Many of the mistakes and unpleasant surprises that tripped up Vogtle should be resolved once a few AP-1000s have been completed. There won’t be another licensing delay while the NRC ponders the shield wall. The builders now know how to do the base mat rebar. Lake Charles, GPSC staff monitors recently noted, has gotten its act together with up-to-snuff submodules and paperwork. Westinghouse engineers will fine-tune a buildable set of blueprints.

The lessons learned at Vogtle and other Gen III+ projects will thus help subsequent builds go smoother, faster and cheaper—maybe even as well as last year’s Gen II projects. But that will only happen if there are subsequent builds. China is planning to build many more AP-1000s along with its own upsized clone, the CAP-1400; that deployment will support a well-oiled supply chain, sub-module factories and an experienced construction workforce. The AP-1000’s future is iffier elsewhere: Westinghouse is pursuing projects in Britain and India, but has no firm orders in the United States. If none materializes, Lake Charles could close, experienced project managers and skilled workers will disperse to other industries, suppliers in the United States will abandon the nuclear sector and perfected blueprints will sit on the shelf.

Avoiding that future will require smart industrial policy. Large-scale, systematic nuclear construction programs run by state-owned utilities have proven their worth in the past in France—still the most successful example of decarbonization—and now China and South Korea. That approach isn’t fashionable anymore in the deregulated electricity markets of the neoliberal West. But less dirigiste industrial policy, in the form of subsidies and mandates, lend crucial support for renewable energy in the United States and Europe. Nuclear power could receive the same supports. One approach being explored in some states is a low-carbon electricity portfolio standard that includes both renewables and nuclear and lets utilities decide what mix of energy sources will best meet decarbonization targets.

Accelerating the global nuclear renaissance will also require a robust international supply chain that lets countries with a thriving domestic industry export their know-how. Modularization, though it faltered at Lake Charles, will be a key element of that, allowing standardized factory production to pre-fab the bulk of a plant instead of inexperienced onsite labor. China will likely be the modular epicenter. Nurtured by a huge domestic deployment, Chinese factories could produce cheap sub-modules and parts for nuclear projects around the world, just as they now make the world’s solar panels. China has also designed a reactor based on an Areva design the Hualong 1, for domestic builds and export. South Korea is ahead of China in nuclear exports; it is supplying four APR-1400 units for the Barakah plant in the United Arab Emirates, now half built. Russia is also building several plants in foreign countries.

Growing an international supply chain for both parts and foreign-designed reactors will require flexibility on nuclear trade and regulation. Domestic-content rules strictures may have to be relaxed. Standards among national nuclear regulators could be harmonized to make it easier for reactor vendors to get licensed in foreign countries, which would spur competition and lower costs. One reason Britain’s Hinkley C project will build the hugely expensive EPR is that it’s the only model currently licensed by the UK’s Office of Nuclear Regulation; the AP-1000 and the ABWR aren’t yet approved even though they have already been licensed by the gold-standard US Nuclear Regulatory Commission. The South Koreans have applied for a US license for the APR-1400, which beat out the EPR for the Barakah project; it could make an attractive option for cost-conscious nuclear projects in the West.

Another much-disputed but valuable principle—bigger is better—should inform nuclear planning. The advantages of size accrue to whole plants as well as individual reactors. Studies find that multi-reactor builds see sizeable declines in construction costs on later units. Big plants also have lower operating costs because payroll and overhead get spread over more terawatt-hours. The RE Ginna nuclear station -a small, money-losing plant in upstate New York that’s likely to close soon- has a 581-megawatt reactor and 600 workers. That is twice the staffing-to-output ratio of New York’s profitable Indian Point plant, which generates 2,070 megawatts from two reactors with a staff of 1,050. Inefficiencies like that help push Ginna’s production costs to 5.6 cents per kilowatt-hour, twice the average for US nuclear plants.34 Economies of scale are why China and South Korea are going big, building enormous plants with up to six gigawatt-size reactors apiece. Construction projects on that scale are hard for private utilities to finance—another reason to try public financing and ownership in an era of cheap government borrowing costs.

The most important lesson to take from last year’s accomplishments in the nuclear industry is that it’s not really what you build, it’s how you build it. Gen II reactors have experienced many epic cost blow-outs, delays and cancellations, but 2015 showed (once again) that they can be built cheap and fast with help from supportive industrial policies. Those policies can benefit the new Gen III models as well. Cheap nuclear power doesn’t require technological leaps, just steady optimization of the designs we have.

1. IAEA PRIS database There were also nuclear closures: six small Japanese reactors that had been shut down for years were officially retired; two operating plants in Britain and Germany with a capacity of 1.77 GW also closed, but their loss was balanced by the restart of two Japanese reactors with a capacity of 1.69 MW and net uprates of 0.46 GW.

2. World Nuclear Association country reports.

3. Bloomberg,


5. “BP Statistical Review of World Energy 2015” workbook. Wind ended 2013 with 321 GW capacity worldwide and ended 2014 with 373 GW, during 2014 global wind generation was 706.2 TWh, for a capacity factor of 23.2 percent. I assume 25 percent because of a higher percentage of offshore wind in 2015’s additions.

6. “BP Statistical Review of World Energy 2015” workbook. Following procedure in note 5 gives a global solar capacity factor of 13.5 percent, I assume 15 percent for last year’s solar installations.

7. For Russia’s Beloyarsk 4 fast reactor I assume the same lifetime capacity factor as its predecessor Beloyarsk 3 fast reactor, 75 percent, from IAEA PRIS database

8. World Nuclear Association China report gives a highest estimate of CPR-1000 costs of $2,300 per kilowatt, assumed to be overnight costs . Financing costs: Rosner and Goldberg, U. of Chicago, “Analysis of GW-Scale Overnight Capital Costs,” specifically their analysis of the original budget for VC Summer in Table 6. They calculate that total project costs after escalation and financing were 36 percent higher than the overnight cost. With an overnight cost of $2,300 per kw for the Chinese reactors, adding $800 per kw for escalation and financing gives a total cost of $3,100 per kw, 35 percent higher than the overnight cost and in line with Rosner and Goldberg’s ratio for VC Summer. South Korea’s Shin Wolsong 2 reactor part of a 2-unit plant costing $4.58 billion or $2,385 per KW, similarly assumed financing costs to $3,100 per KW. Beloyarsk 4, 789 MW, had a total cost of 146 billion rubles, assumed pre-devaluation exchange rate of 30 rubles to the dollar.

10. Bloomberg

11. Bloomberg puts total investment in solar power in 2015 at $161.5 billion, but that seems high compared to press reports of project costs. The IEA report “Projected Costs of Generating Electricity,” 2015 edition puts average utility-scale solar overnight costs at $1,562 per KW; I’ve used that figure for all solar gigawatts in 2015, undoubtedly an underestimate.

12. $3,000 per kilowatt for the 8.6 GW of Chinese and South Korean capacity, plus $4.8 billion for the 789-megawatt Russian fast reactor.

13. Chinese nuclear capacity added in 2015 from IAEA PRIS databse Chinese wind and solar capacity added  Costs of Chinese wind put by Bloomberg at about $1369 per kilowatt. .  Solar installations from press reports. Chinese solar capacity factor from “BP Statistical Review of World Energy 2015” workbook 2014 statistics. Wind capacity factor of 22 percent: In 2015 China generated 185.1 terawatt-hours of wind energy, , from an average grid-connected capacity of 112 GW, for a capacity factor of 19 percent. But 15 percent of wind output was curtailed, without curtailment the wind capacity factor would have been 22 percent

14. China’s 2015 wind generation was 185.1 TWh  up from 153.4 TWh in 2014 Nuclear generation in 2015 was 171 Twh in 2015, up from 133 in 2014








22. IAEA PRIS database,





27. Information on the Vogtle build is drawn from the multiple testimonies of Georgia Public Service Commission oversight staff William Jacobs, Stephen Roetger and Philip Hayet, see GPSC docket 29849. .

28. Testimony of William R. Jacobs, Jr. to Georgia Public Service Commission, December 7, 2012, pp. 16-17.

29. Fourteenth Semi-Annual Construction Monitoring Report for Plant Vogtle Units 3 and 4

30. Testimony of William R. Jacobs, Jr. to Georgia Public Service Commission, December 7, 2012, pp. 16-17. pp.28-29.





36.  , .


38. ;  ;

39. Testimony of of Georgia Public Service Commission oversight staff William Jacobs and Stephen Roetger, December 2012, pp. 20-21, GPSC docket 29849.

40. For Example, ibid; pp. 25-6.

41. Georgia Power 2015 Annual Report, p. 8