Thirty years ago, in March 1979, a group of badly trained operators in the control room at Three Mile Island's unit 2 confronted a minor malfunction. The problem, a simple pump shutdown, was quickly made worse by an instrument panel that failed to inform the operators about a stuck valve and by an alarm system that overloaded after the first malfunction. The operators botched an attempt to solve the rapidly escalating problem, allowing a small leak to drain most of the cooling water out of the $700 million reactor. In about two hours, they converted America's newest nuclear plant, which had begun commercial operation just three months earlier, into a $1 billion liability.
The event at the reactor, near Harrisburg, PA, provoked near-panic, and although government reports said the maximum possible radiation exposure was too small to have much effect on human health, one major casualty was the outlook for the nuclear industry itself. The meltdown did not end the first round of nuclear construction in this country; 50 reactors already under construction were completed after the accident, and orders for new plants had effectively ceased anyway. (The last order for a nuclear plant that was actually built came in 1973.) But for years to come, it remained unthinkable to plan new reactors as part of the nation's energy portfolio.
Given pressures to reduce carbon dioxide emissions from fossil-fuel power plants, however, construction of nuclear plants could be poised to begin anew. The technology has grown more reliable and more efficient. Reactors now run 90 percent of the hours in a year, compared with less than 60 percent in 1979, effectively cutting the capital cost of a kilowatt-hour by about a third. Meanwhile, other sources of power have started looking a lot worse. Congress seems likely to put some kind of price tag on carbon dioxide emissions, so the price of coal-produced electricity could rise by 30 to 50 percent. The price of natural gas is low right now but has been more volatile than the price of oil in the past few months as surging supplies and lackluster demand play leapfrog. Such volatility makes electric companies reluctant to rely heavily on gas.
All the same, the nuclear industry faces tremendous risks, though their nature has changed since 1979. As the possibility of an accident that panics or injures the neighbors has diminished, the likelihood has grown that even a properly functioning new reactor will be unable to pay for itself. And changes in the utility industry since 1979 mean that this time, the money a company wastes may be its own.
Whether new nuclear plants are a good bet economically depends on three factors, all now in flux. First is the cost of a new reactor. In 2005, a few would-be reactor builders said they could construct a facility generating 1.2 to 1.6 gigawatts for $2,000 per kilowatt of capacity. Now, they put the cost at $4,000 per kilowatt. Neither price includes interest charges accrued during construction, which could be substantial if the job takes more than the five years or so that the builders predict--or if interest rates rise, as they are expected to. The Electric Power Research Institute, a utility consortium based in Palo Alto, CA, recently put the capital cost of a new coal plant at under $3,000 per kilowatt and that of a natural-gas plant at $800 per kilowatt.
The second factor is uncertainty about possible future competitors. If 10 years from now wind or solar plants, or coal plants that capture their carbon emissions, are able to deliver vast amounts of cheap power, the market price of electricity will fall, and plant owners may never see enough revenue to meet their costs.The third factor is uncertainty about the price of fossil fuels, particularly natural gas. In the last year, the fuel cost for a kilowatt-hour generated from natural gas has varied from about 2.3 cents to about 9 cents. If a federal cap-and-trade system or a tax on carbon dioxide emissions is instituted, that is likely to add 0.5 to 1.5 cents per kilowatt-hour. Add in 2 cents or more to recover the cost of building the plant, and the price of gas-fired power could make nuclear power look very attractive--or really overpriced.
A power-producing company that bets on natural gas can choose the size of its wager: a 100-megawatt plant, or a 500-megawatt or 1,500-megawatt one. Conventional nuclear plants come in only one size: jumbo. Some power companies have proposed smaller plants, but costs for factors like labor and security are mostly insensitive to size, so these costs per kilowatt-hour rise as the plant shrinks. Costs for engineering and materials are also greater per kilowatt-hour the smaller the plant is.
All these economic risks matter for nuclear power now, because the electricity marketplace has changed dramatically since the industry was deregulated in the 1990s. Before that, each plant's output was paid for by consumers, no matter what the cost. As a result, millions of consumers got stuck paying more than they should have, because their local utilities unwisely chose nuclear instead of coal or natural gas. The financial rules differed from state to state, but generally, once a plant was in service, a company could collect a specified return on its investment, and if a plant projected to cost $1 billion ended up costing $2 billion, the customers paid.
In today's electricity market, however, producers in many states are paid according to market price. Companies build a plant for whatever price they can manage and sell electricity for whatever price they can get. If a reactor produces power at 10 cents per kilowatt-hour and a natural-gas plant produces it at 12 cents, the reactor builder makes a killing. Reverse the numbers and the reactor builder gets killed.
The electricity industry won't build much of anything these days without government help, in the form of loan guarantees, production tax credits, guaranteed markets, or, preferably, all three. Wind now gets bigger production subsidies than nuclear on every kilowatt-hour generated, proportionally more loan guarantees, and a guaranteed market: many states insist on a certain quota of renewable energy, sometimes regardless of cost. In contrast, nuclear power receives production subsidies on only the first 6,000 megawatts of capacity (four or five reactors' output), and its pool of loan guarantees is shrinking relative to the price of construction.
"Right now, the federal incentives are much more conducive to pushing forward renewables," said Jim Miller, the chief executive of the energy company PPL, in June. His company, based in Allentown, PA, would like to build a reactor but will not do so without federal loan guarantees. It will not get them, at least not under the 2005 Energy Policy Act, in which Congress approved only enough to assist a handful of plants: $18.5 billion. "Nothing is currently in place to move the nuclear industry along at the pace people perceived it would move when the 2005 act was passed," Miller says.
The idea of the legislation was that Congress would spoon-feed financial aid to the first half-dozen or so new nuclear plants, and others would follow on their own once new designs were demonstrated and a reformed licensing process was in place. Now, it looks as if those half-dozen new reactors will be the limit of the "renaissance," unless more help is forthcoming. The industry lacks the votes in Congress to expand the loan-guarantee program. Subsidies for wind and solar power are popular, in part because they can be justified as aid to emerging technologies. But many legislators feel that nuclear is less deserving of taxpayer support.
Even now, nuclear power has the potential to be economically attractive if costs and competition are favorable--and if overall demand for power remain strong, with high industrial use and limited improvements in efficiency.
All of that is possible. But the odds are probably not good enough for the nuclear industry to place a bet with its own money. Only the government can agree to back up that bet, and it has yet to do so.
Matthew L. Wald is a reporter at The New York Times. His feature "The Best Nuclear Option" appeared in the July/August 2006 issue of Technology Review.
By Matthew L. Wald