The following is a contributed article by John Reilly, a co-director of the Massachusetts Institute of Technology Joint Program on the Science and Policy of Global Change, and an energy, environmental and agricultural economist.
If solar panels and wind turbines keep getting cheaper, why bother building anything else?
Because as we add more solar panels and wind farms, their productivity declines. And while the cost of individual solar panels is low, when there are enough of them, they impose real costs on the rest of the system.
A new study from a team of researchers at MIT (including myself) examines these trends and explains why this creates an important role for both existing and new nuclear power plants in an affordable decarbonized energy system.
Wind and solar costs
There is still lots of room for solar and wind to grow in the marketplace, and they should grow many-fold; however, our study found that the costs they impose on the system begin to become substantial once the variable renewables (wind and solar) provide about 40% of the total electricity generation in the region — about four times larger than what those sources provide today for the country as a whole.
But in some regions, the market penetration of wind and solar is far higher, and acute problems are emerging. And 40% is well below the vision of a grid powered solely by wind and solar that some advocates call for.
Raising the share of electricity produced by sun and wind above 40% creates at least a couple of adverse effects on the electricity system: First, more and more solar and wind production is "curtailed" — that is, the generator must be unplugged from the grid during its most productive hours because more electricity is being produced than is needed.
Second, more back-up generating capacity is needed to fill in when the wind and solar are not producing. Since that extra backup capacity is idle much of the time, it adds costs to the system.
If the electric system has to run 24/7, through periods of high and low wind and sun production, and of high and low electric demand, then integrating increasing amounts of intermittent renewables becomes progressively more difficult.
California is at the leading edge of deploying intermittent renewables. In April of this year, the state unplugged renewable generators that would have produced 190 gigawatt-hours, worth tens of millions of dollars at average U.S. retail prices, though it was worth nothing at the hours when it would have been produced.
On May 27 at 1 p.m., operators shut down 4,700 megawatts of renewable capacity, a record. Most of it was solar. For comparison, the Hoover Dam is 2,000 megawatts.
Storage can help, but storage is mostly an hour-to-hour or day-to-day strategy, and even that is more difficult than it looks, when you study the problem in detail.
Battery manufacturers keep their pricing data murky, but a lithium-ion battery that will hold one kilowatt-hour goes for about $300. That's less than it used to be, but too expensive for large-scale storage of a commodity that sells for about 12 cents.
And while it may someday make sense for daily storage to run a household all evening on noontime sun, it's hard to imagine storing the energy from March winds for the still nights of August, or storing June sunshine for the still, dark days of January.
Nuclear and carbon capture
As a result, to affordably achieve a 90% reduction in carbon emissions from the electric sector by mid-century compared to 2005 levels, the system will need other tools, notably nuclear energy and possibly carbon capture. (But we've got a lot more work to do before carbon capture is inexpensive enough for mainstream use.)
Nuclear energy could be very useful even if its cost per kilowatt-hour is higher than solar or wind, because it will deliver that energy at times when renewables are not available. It doesn't matter how cheap renewables are if they cannot generate power when it is needed.
But for nuclear energy to play a substantial role in decarbonization, the study finds, it will require finding a way to cut the cost of building new plants, at least modestly.
This won't happen quickly. Nuclear construction goes smoothly when the builders are experienced and proficient, and we're out of practice. And building a "first-of-a-kind" anything is slow and painful. They don't call it a learning curve for nothing.
But the energy industry shows that hardware and human performance do follow a learning curve. Witness wind farms, or fracking.
Too many eggs in one basket
The essential problem with putting too many eggs in the wind and solar basket is that the strategy requires lots more generating capacity to meet a given level of load.
In the old days, when the fossil fuel plant started up whenever you needed it, all that was needed was sufficient capacity to meet demand, plus a margin to cover possible breakdowns; now, you need enough capacity to balance the loss of generation capacity from solar and wind at times.
A "100-megawatt" solar farm only generates 100 megawatts of power at mid-day; peak demand is usually many hours later, when the sun is low in the sky and the solar field is producing only a fraction of its capacity. Same story for wind, which usually generates power at its rated maximum capacity at night, when demand is low; when the world is awake and demanding electricity, the wind turbines are producing only a fraction of their potential maximum.
So if a solar panel generates at, say, 50% of capacity during peak demand periods, and a wind turbine at 20%, then even if the cost of the hardware has fallen, more and more capacity is needed to keep the lights on. A renewable-centric system that is large enough to meet peak demand would inevitably face periods where it generates a lot of surplus power, which has no value.
By 2050, the useless surplus could be 20% of the amount that wind and sun produce, if we aim, as we must, to decarbonize the electricity sector.
As the productivity of solar panels and wind turbines fall because of curtailments, and the integration costs rise, there is a need for other zero-carbon generators that are available at all times, such as nuclear energy.
Modest reductions in the cost of new nuclear power would create a substantial role for new reactors, and sharply reduce the need for a carbon price to reduce carbon output.