Dollars, Sense, and Kilowatt-Hours

"Levelized Cost of Energy" has long been used to generate misleading comparisons of nuclear and renewables. Here's why even former proponents of LCOE are moving on.

Dollars, Sense, and Kilowatt-Hours

Fans of renewables have long asserted that electricity captured right from the wind and sun is many multiples less expensive than electricity from new nuclear plants—as demonstrated by calculations of something called the Levelized Cost of Electricity. The most widely quoted LCOE is the estimate Lazard, the French asset management and advisory firm, puts out every year.

But Lazard is having second thoughts.

When the company issued its most recent annual analysis of LCOE, it stated firmly that this analytical tool has been widely misused to compare apples and oranges. The refutation is reminiscent of the scene in the movie “Annie Hall,” in which Woody Allen, irked by an academic’s faulty reasoning about the work of media guru Marshall McLuhan, drags McLuhan himself out of a corner and watches as the famous theorist tears the academic apart.

LCOE has been useful for a long time, but it was never intended to deal with intermittent energy sources. Essentially, it calculates what it costs to build and run a power plant and then figures out how much electricity it will be called upon to produce over its lifetime. It divides the former by the latter, and then comes up with a single number that can be used to compare coal, gas, hydro, and nuclear power facilities.

The problem is that the underlying assumption of the LCOE calculus—so obvious that it was seldom stated—was that any given power plant would run when needed, and not run when not needed, and thus that the production would have value. In the days before solar and wind, that assumption was correct. A utility could calculate how many hours a year it was likely to need a gas peaking plant, or a baseload coal plant.

But the technique doesn’t take account of the fact that a kilowatt-hour varies in value according to market conditions, a problem for generators that can’t choose when they run, like wind and solar, which run when mother nature says it is time. Wind, and especially solar, have fratricidal tendencies; that is, when they are at maximum production, they tend to flood the market and push down prices. During the best production times, they are bleeding each other, economically.

Two years ago, the Breakthrough Institute published research on this problem. And BTI isn’t alone. The Electric Power Research Institute, a non-profit consortium of utilities, pointed out the problem in a 2022 report. The tone is characteristic of engineers addressing accountants:

“Levelized-cost metrics are incomplete for evaluating the relative competitiveness of system resources. Comparing traditional levelized-cost metrics between different resources with dissimilar characteristics (e.g., contrasting the levelized cost of electricity for solar and nuclear power) ignores differences in the value of the resources to the system.” To determine the value of an energy resource—value, as distinct from cost—requires modeling the interactions that occur within the system, EPRI said.

And as Robert Idel, an energy economist at Rice University, has pointed out, the purpose of an electricity system is not to generate electricity; it is to “supply a specified amount of electricity to a particular place at a specific time.” Economically, “the fact that intermittent generation has no obligation to meet the demand can be seen as a hidden subsidy,” he argues. The problem with such subsidies will become more obvious as dispatchable generation—that is, gas, nuclear, and coal facilities that can be turned on and off as demand indicates—is pushed out of the market in favor of wind and solar, and there is no one left to fill in the gaps.

Idel, instead, proposes an alternate measure, Levelized Full System Cost of Energy, which includes factors like transmission and storage. Idel suggests calculating what the cost would be if the renewable technology were called upon to supply 95 percent or 100 percent of demand, an exercise that “challenges the economic sanity of 100 percent intermittent renewable targets.” But to the public and elected policymakers, LCOE is appealing because it is straightforward and “catchy,” despite being misleading, he points out.

Lazard’s own latest analysis stresses a new way to look at prices, including the cost of “firming” the intermittent sources, by having enough dispatchable capacity on hand to assure adequate supply. Pinpointing that price is tricky, because in most parts of the country, power plants earn their keep by making electricity for lots of hours. Pre-solar, a gas-fired power plant would run during all the daylight hours, and now it may only run when there isn’t a lot of sun. With its costs spread over fewer kilowatt-hours of production, each kilowatt-hour will cost more.

Renewable energy, without subsidies, is competitive with conventional generation “under certain circumstances,” the study finds. But not the type that is politically most popular – rooftop solar.

The most expensive conventional generation is up to $221 per megawatt-hour for gas peaking plants. That’s the same price for nuclear, though coal and gas combined-cycle are less expensive.

Putting solar on the roof of a house, as many like to do, is in the same range: $117 to $282 per megawatt-hour, according to Lazard. The way to build solar economically—in a way that actually competes on price with conventional generation, the report suggests—is at utility scale, which usually means bulldozing a patch of land and installing panels on the ground. That produces electricity at $24 to $96 per megawatt-hour.

(Prices differ partly because some projects are in sunny deserts. Rooftop solar is more expensive partly because each project has to be engineered and installed separately. And the locations aren’t always particularly sunny. But California requires them on new houses, and the federal government and many states subsidize them heavily, because they are popular with voters.)

The cost of “firming,” too, rises as penetration rises. Making up for an intermittent source that provides 1 percent of your energy is easy; California is now at 32 percent solar, and that’s a problem. With “firming” costs included, California solar that started out at $43 per megawatt-hour jumps to $141, on an unsubsidized basis. And when backed up by gas, it’s no longer carbon-free.

Solar with storage is $117, Lazard says and engineers will argue over how much storage is needed. It varies by location and other factors. Stored energy from solar also has a larger carbon footprint; each kilowatt-hour has a carbon footprint, and so do the batteries, and the batteries are like leaky buckets, giving back less than they take in.

These cost numbers are all based on assumptions for interest rates and subsidy policies. If those change, along with the price of gas or coal, or the price charged for carbon pollution, that re-arranges all the numbers. Lazard also describes a trend that indicates that the LCOE of solar, long dropping, is now rising. According to Lazard, the price of a megawatt-hour from solar dropped from $359 in 2009 to $38 in 2021, but is now back up to $60. Nuclear, based on the price of the new Vogtle reactors, is at $180.

The capital cost, per kilowatt of capacity, for zero-carbon technologies ranges from $700-$1,400 for utility-scale solar to $8,475 to $13,925 for nuclear. Nuclear capacity has a far higher output, because it runs nearly all the hours of the year, but a key goal of advanced nuclear is to bring the capital cost down.

The bottom line is that it doesn’t matter what a technology costs; it matters what the electricity system needs. There is a least-cost combination of technologies that are required to serve load, but you don’t serve load by picking only the least expensive generator. The generators have other attributes, including cleanliness, reliability and dispatchability, to consider. What matters isn’t their cost, it’s their value.