Nuclear proponents are launching a full-court press for fresh investment in the technology. The release of the new film Pandora’s Promise, another editorial from ardent nuclear champions Michael Shellenberger and Ted Nordhaus of the Breakthrough Institute, and Paul Blustein’s recent piece in Quartz, “Everything you thought you knew about the risks of nuclear energy is wrong,” are part of an effort to put a new shine on a technology that once offered, but failed to deliver, electricity “too cheap to meter.”
All of these actors are purportedly motivated to support nuclear power on climate grounds, emphasizing the technology’s extraordinarily small physical footprint, its ability to generate massive amounts of electricity, and its lack of carbon emissions (after the plants are built). And they are probably right that the risks of radiation have been historically overblown as “junk science” wormed its way into popular culture. But the anti-nuclear crowd (and I, too, used to count myself among them) is probably right for the wrong reasons.
Missing from the entire debate about nuclear is the most important fact of all: Nuclear is dying due to poor economics, and the debate is already over as far as the market is concerned.
Shellenberger and Nordhaus have backed up their arguments with junk accounting on nuclear energy’s costs. This is where the discussion must depart from mere boosterism and descend into the deep, dark world of energy economics—a subject that Blustein did not even address.
The generally accepted way to compare the cost of various power generation technologies is a levelized cost of energy (LCOE) analysis. There are valid questions about this approach, which serious energy analysts continue to wrangle over. But for mere mortals and policy advocates, LCOE will have to do. There isn’t a better alternative.
As the Energy Information Administration (EIA) explained in the LCOE analysis section of the Annual Energy Outlook 2013:
[LCOE] represents the per-kilowatthour cost (in real dollars) of building and operating a generating plant over an assumed financial life and duty cycle. Key inputs to calculating levelized costs include overnight capital costs, fuel costs, fixed and variable operations and maintenance (O&M) costs, financing costs, and an assumed utilization rate for each plant type. The importance of the factors varies among the technologies. For technologies such as solar and wind generation that have no fuel costs and relatively small O&M costs, the levelized cost changes in rough proportion to the estimated overnight capital cost of generation capacity. For technologies with significant fuel cost, both fuel cost and overnight cost estimates significantly affect the levelized cost. The availability of various incentives, including state or federal tax credits, can also impact the calculation of levelized cost. …As with any projection, there is uncertainty about all of these factors and their values can vary regionally and across time as technologies evolve and fuel prices change.
Anyone who really wants to understand the costs of power generation should read that report, as it explains the many factors, assumptions and uncertainties that a good LCOE analysis entails.
In the EIA’s analysis, which leaves out all incentives, the average cost of “advanced nuclear” or “next-generation nuclear” plants entering service in 2018—long lead times associated with these technologies will make it difficult to open any early—would be $108.40 per megawatt-hour (MWh), equivalent to $0.1084 per kilowatt-hour (kWh), in 2011 dollars. This seems in the right ballpark, as the estimated cost of power from the new nuclear plant under construction in the Kaliningrad region of Russia is around $0.10/kWh, a German lawmaker said in April.
For reference, the 2012 average retail price of electricity in the US was $0.1153/kWh. So the cost of new advanced nuclear power would be just barely below the retail price of electricity—power sold to you and me at home. (Commercial, industrial, and transportation customers all buy power for less than the LCOE cost for advanced nuclear power.)
In other words, it would be very difficult for a utility to make money selling power generated by advanced nuclear plants, if they had to shoulder the entire cost themselves. But they don’t.
Not included in the LCOE analysis is the cost of decommissioning nuclear plants, which is often externalized and pushed onto ratepayers through surcharges on their utility bills, or the cost of managing nuclear waste for decades, which is generally pushed onto taxpayers through the Department of Energy budget. And these are not trivial costs: Edison International estimates that decommissioning its San Onofre Nuclear Generating Station near San Diego, which it permanently retired last week, will cost around $3 billion. So the LCOE analysis actually understates the true, all-in cost of nuclear power.
But the complexity doesn’t end there. As EIA explains, the true cost of power generation can vary substantially based on a number of other factors specific to where the plant is located, including the utilization rate (which depends on the demand character and the existing resource mix where the plant is located), the existing mix of resources in the area, the capacity value (how much of the time the plant will run) on the local grid, and the portfolio diversification needs (the specific mix of generation technologies) of the local utility.
Recognizing these factors, EIA suggests a minimum cost for advanced nuclear of $104.40, an average of $108.40, and a maximum of $115.30/MWh.
Solar photovoltaics (PV) in 2018 would range from a minimum of $112.50, to an average of $144.30, to a maximum of $224.40/MWh.
Wind energy would range from a minimum of $73.50, to an average of $86.60, to a maximum of $99.80/MWh.
The Breakthrough Institute points to this same EIA analysis as proof that “solar costs substantially more than new nuclear construction,” which is correct if one only looks at the average prices. But it’s more complicated than that.
The EIA has historically overestimated the cost of renewables, and underestimated the cost of conventional fuels. The new 50-MW Macho Springs solar plant under construction by First Solar in New Mexico is will deliver power for $50.79/MWh under its Power Purchase Agreement (PPA), and other US solar projects have come in this year in the range of $70 to $90/MWh.
By those recent numbers, the cost of US solar PV is already as little as half that of advanced nuclear generation in 2018. Further, we should bear in mind that the cost of solar and wind is still falling, while the cost of nuclear keeps rising.
Comparing the cost of one power generation source to another can get much more complex still, including the crucial matter of commodity and construction costs between now and 2018—a subject fraught with uncertainty all on its own—and future policy decisions.
The BTI’s duff maths
The Breakthrough Institute elects to ignore all of this real-world complexity and offer its own extremely distorted way of comparing power generation costs.
In its most recent analysis, “Cost of German Solar Is Four Times Finnish Nuclear”, BTI compares the retail cost of German solar, which includes significant feed-in tariff incentives, to the capital cost of building a new nuclear plant in Finland, as estimated by its developer, plus the EIA’s estimate for the fixed and variable costs of nuclear power. There are numerous glaring problems with this approach, but I’ll name just the obvious ones.
It adds up the cost of all installed German solar PV from 2000-2011, a period in which the price of solar fell dramatically, and a fact that Shellenberger and Nordhaus even recognized. Thus it is weighted to the much higher costs of the past decade, rather than current costs, let alone the cost of PV in 2016 when the Finnish nuclear plant isexpected to enter service. An analysis based on the actual, unsubsidized cost of German solar PV in 2016 would find that it is below the cost of new nuclear power, not four times as expensive. Even the subsidized cost would be lower. As Craig Morris pointed out last week in Renewables International, the lowest solar PV feed-in tariff ($0.13/kWh) on Germany’s sliding scale is already below the cost of EDF’s proposed new nuclear plant in the UK ($0.15/kWh), and one year from now, Germany’s highest solar feed-in tariff will be too, at $0.13/kWh.
It’s based on the post-incentive cost of German solar, not the cost of the technology. At $0.1125/kWh in the EIA’s LCOE analysis, the minimum cost of solar (which is well above recent US solar PPA contracts), without incentives, is below the anticipated cost of EDF’s nuclear plant in the UK.
It uses the developer’s latest cost estimate for Finland’s Olkiluoto 3 reactor, which Shellenberger and Nordhaus note is seven years behind schedule and nearly three times over its initial budget. Developer cost estimates should always be viewed with skepticism; an analysis by the Congressional Budget Office, cited in a 2009 analysis by the Union of Concerned Scientists, found that utility estimates for nuclear plant costs are usually around one-half the estimates of independent analysts and Wall Street. In my experience, the estimates of the latter group are usually close to the mark. Until the final costs are known, the estimates BTI cites for nuclear plants under construction in France and China are no more credible.
The PPA costs cited above for current PV installations in Germany and the US are rock-solid, contractually guaranteed prices, with no externalities. BTI is citing squishy estimates for the “overnight capital cost” (without factoring in interest rates) for a nuclear plant that won’t even be completed until 2016. Which set of estimates would you consider more credible?
Over the years I have spent many, many hours reviewing the cost estimates for new nuclear plants. Every time I’ve gone down that rabbit hole I have reached a point where I threw up my hands and quit because the data quality is so poor. Since no nuclear plants have been built on schedule in recent years, there are no reliable real-world cost data to establish a baseline. When you explore the various components of an estimate, you quickly find yourself in a regression of footnotes to previous papers which lead you back to estimates made a decade or more ago, before the cost of all commodities exploded in the second half of the 2000s.
In short: Cost estimates for new nuclear plants are not credible. I have yet to find a single one that stood up to close scrutiny. And as far as I am aware, no nuclear plant has ever been built for close to its original cost estimate.
With numerous, highly transparent LCOE analyses available from EIA, NREL, and other agencies, why does the BTI ignore them in favor of their own, partial analysis, based on a developer’s cost estimate?
They appear to have begun with the predetermination that nuclear power is the only solution to everything, and then rounded up a highly selective, distorted, and outright wrong pile of evidence to make their case.
Nuclear’s long goodbye
The simple fact is that, at least in the US, the nuclear industry is dying a slow death. The announced closure of four major facilities in 2013 alone amount to 4,246 megawatts of nuclear capacity—enough to power 2.7 million homes for a year—that are being retired.
Even while the nuclear industry is able to externalize its costs for insurance (which are federally limited), loan guarantees (which are federally backstopped), decommissioning (which is pushed onto ratepayers) and waste handling (which is pushed onto taxpayers), it still lost. If it had to stand on its own and pay its full insurance costs like every other energy source, we could never build another nuclear plant in America, because no private investors would be willing to take that kind of risk. It’s hard to imagine how the economics could be more tilted in nuclear’s favor (although I’m sure its proponents have ideas on that).
The reason nuclear is dying is economics, not tribalism, as Shellenberger and Nordhaus claim. The UCS study found that if the EIA’s National Energy Modeling System were updated using 2009 utility cost estimates for building new nuclear plants, instead of EIA’s old over-optimistic projections, nuclear plants would not be the most economical way to add new capacity. The economics have shifted even farther away from nuclear since then.
Meanwhile, outside the fantasy world inhabited by the BTI, the real energy market has moved on. The US installed 13,200 megawatts of wind capacity in 2012, according to Bloomberg New Energy Finance.
In 2010, the combined generation of the four nuclear plants now headed for retirement, would have been just over 30 million MWh if all four were operating normally (one plant, the Crystal River plant in Florida, was not). Using a weighted average of the EIA’s onshore and offshore capacity factors for wind (36%), the wind capacity installed in the US last year will generate 41.6 million MWh per year.
In other words, the US installed enough wind last year to replace 138% of the nuclear generation shuttered this year.
The US nuclear fleet is shrinking, while wind and solar are posting double-digit, exponential growth rates. What part of this do Shellenberger and Nordhaus not understand?
To be sure, there are other important considerations from a policy standpoint. Insurance costs for nuclear power are probably inflated, due to exaggerated fears about radiation. But nuclear advocates aren’t arguing that underwriting rules should be updated. Yes, the utility industry will need to update the century-old hub-and-spoke architecture of the grid to accommodate high levels of variable and distributed renewable power, but that needed to be done anyway. Yes, there are important questions to be addressed about dispatchability—the degree to which the power generator can be fired up at will, as needed—voltage regulation, service level guarantees, and the evolution of utility business models and regulatory environment, which I have written a series of articles about (see here, here, here, and here).
But it is simply wrong to claim that solar PV is four times as expensive as nuclear power, in Germany, Finland, or anywhere else. BTI’s entire argument—including its repeated insistence that we should emphasize innovation over deployment of renewables—is bizarre, and detached from reality.
These are the facts: Renewables have taken the lead in new power generation in America, comprising nearly half of all new generating capacity installed in the United States in 2012. In the first quarter of this year, nearly half the new capacity installed was solar. With its poor economics, enormous complexity, overly-large capital requirements, too-long lead times, and overall risk, US nuclear power is headed for contraction, not resurgence. Ultimately, I think the same will be true globally.
I share BTI’s passion for creating a low-carbon future to address the threat of climate change. But false accounting isn’t the way to go about it. We should lean hard on solutions that work today, like renewables, demand response, microgrids, and advanced grid management technologies.
There may come a day when next-generation nuclear reactors can prove their economic viability in the real world and not just on paper. But until they do so—and especially while the old generation of nuclear power is dying—it makes no sense to promote nuclear power.
More from Quartz